Biotehniška fakulteta Univerze v Ljubljani Biotechnical Faculty University of Ljubljana AACTA AGRICULTURAE SLOVENICA96•22010 Acta agriculturae Slovenica • ISSN 1581-9175 • 96 – 2 • Ljubljana, december 2010 Acta agriculturae Slovenica Volume / Letnik 96 · Number / Številka 2 · 2010 VSEBINA / CONTENTS Peter DOVČ 67 Prof. ddr. Franc LOČNIŠKAR, dipl. inž. agr. (1923–2010) In Memoriam Prof. Dr. Franc LOČNIŠKAR, Ph.D. (1923–2010) GENETIKA / GENETICS Gregor GORJANC, Tina FLISAR, Jose Carlos MARTÍNEZ-ÁVILA, Luis Alberto GARCÍA-CORTÉS 69 Simple reparameterization to improve convergence in linear mixed models Enostavna reparametrizacija za izboljšanje konvergence linearnih mešanih modelov BIODIVERZITETA / BIODIVERSITY Abdulmojeed YAKUBU, Kingsley Omogiade IDAHOR, Hadiza Salihu HARUNA, Matthew WHETO, Samuel AMUSAN 75 Multivariate analysis of phenotypic differentiation in Bunaji and Sokoto Gudali cattle Multivariatna analiza fenotipskih razlik med “Bunaji” in “Sokoto Gudali” govedom MIKROBIOLOGIJA / MICROBIOLOGY Blaž STRES 81 Antibiotic-resistant soil bacteria in high-altitude (5000–6000 m) soils of the Himalaya Na antibiotike odporne bakterije v visokogorskih tleh Himalaje (5000–6000 m) Maša VODOVNIK, Mirjana BISTAN, Maša ZOREC, Romana MARINŠEK LOGAR 87 Methylmercury inhibits growth and induces membrane changes in Pseudomonas putida Metil živo srebro inhibira rast in povzroča spremembe v membranah bakterije Pseudomonas putida PREHRANA / NUTRITION Monika MARIN, Tomaž POLAK, Lea GAŠPERLIN, Božidar ŽLENDER 95 Variations in the fatty acid composition and nutritional value of Adriatic sardine (Sardina pilchardus Walb.) through the fishing season Maščobnokislinski profil in prehranska vrednost jadranske sardele (Sardina pilchardus Walb.) v odvisnosti od sezone ulova ETOLOGIJA / ETHOLOGY Manja ZUPAN, Daniela BOJKOVSKI, Ivan ŠTUHEC, Dragomir KOMPAN 103 Foraging behaviour of sheep at pasture with different types of paddock Obnašanje ovc na kraškem pašniku z različno vegetacijo BIOTEHNOLOGIJA / BIOTECHNOLOGY Jurij POHAR, Klavdija STRGAR 111 Fluctuating asymmetry in diploid female and sterile triploid rainbow trout (Oncorhynchus mykiss) Fluktuacijska asimetrija pri diploidnih in sterilnih triploidnih samicah kalifornijske postrvi (Oncorhynchus mykiss) ŽIVINOREJA / ANIMAL BREEDING Martina PLANINC, Janez RUS, Milena KOVAČ, Špela MALOVRH 117 Ocena parametrov disperzije za lastnosti zunanjosti pri konjih haflinške pasme Estimation of dispersion parameters for linear type traits in the Haflinger horses Tomaž BARTOL 127 Subject index by Agrovoc descriptors Predmetno kazalo po deskriptorjih Agrovoc Nataša SIARD 129 Subject index by Agris category codes Vsebinsko kazalo po predmetnih kategorijah Agris 131 Abecedno kazalo avtorjev Author’s index 133 Navodila avtorjem 135 Notes for authors Acta argiculturae Slovenica, 96/2, 67–68, Ljubljana 2010 Prof. ddr. Franc LOČNIŠKAR, dipl. inž.agr. (1923–2010) Od prof. ddr. Franca Ločniškarja, zaslužnega profe- sorja Univerze v Ljubljani in častnega člana Slovenskega genetskega društva, smo se poslovili 14. oktobra 2010. Rodil se je 23. aprila 1923 na Turjaku kot peti otrok v učiteljski družini. Po gimnazijskih letih v Zavodu sv. Stanislava v Šentvidu in maturi na Klasični gimnaziji v Ljubljani se je leta 1945 vpisal na Kmetijsko-gozdarsko fakulteto v Zagrebu, kjer je decembra 1949 diplomiral. V tem času je na njegov nadaljnji razvoj pomembno vplival akademik prof. dr. Alojz Tavčar, ki ga je navdušil za ge- netiko, ki je postala osrednja tema njegovega kasnejšega raziskovalnega dela. Prof. Ločniškar je bil prvi doktorand Biotehniške fakultete v Ljubljani (1959), leta 1960 pa je s temo s področja kvantitativne genetike doktoriral tudi na Georg-August Univerzi v Göttingenu v Nemčiji. Kot profesor je vse svoje delovno obdobje deloval na Bioteh- niški fakulteti Univerze v Ljubljani, katere dekan je bil v letih 1979–1981. Univerza v Ljubljani mu je po upokoji- tvi leta 1991 za izjemne zasluge na pedagoškem in raz- iskovalnem področju podelila naziv zaslužni profesor. Za svoje delo je prof. Ločniškar prejel številne nagrade, med drugim tudi nagrado Sklada Borisa Kidriča (1959), Jesenkovo priznanje (1974), medaljo dela z zlatim ven- cem (1979) ter Svečano listino in zlato plaketo Univerze v Ljubljani (1984). Raziskovalno delo prof. Ločniškarja obsega raziska- ve na področju populacijske genetike in genetike kvan- titativnih lastnosti, kjer se je ukvarjal z oceno genetske in okoljske komponente variance in kovariance proizvo- dnih lastnosti. Med prvimi v Evropi je določil dednostni delež za pomembne gospodarske lastnosti prašičev. V se- demdesetih letih prejšnjega stoletja se je začel ukvarjati s citogenetiko. Po prvem opisu kromosomske transloka- cije pri prašiču (1974), je v sodelovanju z I. Gustavsso- nom, M. Hageltornom in L. Zech z Univerze v Uppsali na Švedskem leta 1976 objavil svoj najodmevnejši članek “Cytological origin and points of exchange of a reciprocal chromosome translocation (1p-; 6q+) in domestic pig” v reviji Hereditas. Na tem področju so nato sledili še opisi mozaicizma spolnih kromosomov in avtosomov ter Ro- Acta agriculturae Slovenica, 96/2 – 201068 P. DOVČ bertsonove fuzije pri govedu, izdelava kariotipov postrvi in lipana ter kariotipov za medvrstne hibride. Prof. Ločniškarja so dolga leta raziskovalno zapo- slovali učinki inbridinga in križanja pri selekciji domačih živali. Z izvajanjem dvosmerne selekcije pri kokoših je vzpostavil pomemben genetski model, ki ga raziskovalci na Oddelku za zootehniko Biotehniške fakultete še ve- dno s pridom uporabljajo. Prof. ddr. Franc Ločniškar je tudi pomembno prispeval k razvoju animalne biotehno- logije v Sloveniji. In Memoriam Prof. Dr. Franc LOČNIŠKAR, Ph.D. (1923–2010) Prof. Dr. Franc Ločniškar, Ph.D, a honorary profes- sor of the University of Ljubljana and honorary member of the Slovenian Genetics Society was buried at Ljubljana cemetery Žale on October 14. 2010. He was born on April 23. 1923 in Turjak as a fifth child in a teacher’s family. After his high school years at Diocesan Classical Gymnasium at St. Stanislav’s Institu- tion he entered the study of Agriculture at the Faculty for Agriculture and Forestry in Zagreb in 1945, from which he graduated in 1949. During his study period in Zagreb was of tremendous importance for his further develop- ment his contact to the Academy member Prof. Dr. Alojz Tavčar, who initiated his enthusiasm for Genetics which remained the central topic of his research interest for his entire career. Prof. Ločniškar was the first Ph.D. student who obtained his Ph.D. from the Biotechnical Faculty at the University of Ljubljana in 1959. In 1960 he received his second Ph.D. for his thesis in quantitative genetics from the Georg-August University in Göttingen, Germa- ny. Professor Ločniškar spent his entire teaching career at Biotehcnical Faculty, University of Ljubljana, where he served as a Dean of the Faculty from 1979 to 1981. At the occasion of his retirement in 1991 he received the title of honorary professor of the University of Ljubljana. For his research work he got numerous awards: Boris Kidrič Foundation research award (1959), Jesenko award (1974), State medal with golden wreath (1979) and Gold- en plaque of the University of Ljubljana (1984). Prof. Ločniškar started his research work in the area of population and quantitative genetics with estimations of environmental and genetic components of variances and covariances for production traits. He was among first geneticists in Europe who estimated heritabilities for economically important traits in pigs. In the nineteen seventies, his research interest focused on cytogenetics. After first description of chromosomal translocation in pig (1974), he published in collaboration with I. Gus- tavsson, M. Hageltorn and L. Zech from the University in Uppsala his most important article “Cytological ori- gin and points of exchange of a reciprocal chromosome translocation (1p-; 6q+) in domestic pig” in the journal Hereditas in 1976. In the following years his work was fo- cused on mosaicism of sex chromosomes and autosomes, detection of Robertson’s fusion in cattle and karyotyping of trout, grayling and interspecies hybrids. Prof. Ločniškar also studied effects of inbreeding and specific crossing combinations in selection of domes- tic animals. He established divergently selected chicken and created an important experimental model which is still frequently used by the researchers at the Department of Animal Science. Prof. Ločniškar’s contribution to the development of animal biotechnology in Slovenia was also significant. Prof. dr. Peter Dovč Acta argiculturae Slovenica, 96/2, 69–73, Ljubljana 2010 doi:10.2478/v10014-010-0017-x COBISS: 1.01 Agris category code: U10 SIMPLE REPARAMETERIZATION TO IMPROVE CONVERGENCE IN LINEAR MIXED MODELS Gregor GORJANC 1, Tina FLISAR 2, Jose Carlos MARTÍNEZ-ÁVILA 3, Luis Alberto GARCÍA- CORTÉS 3 Received October 08, 2010; accepted December 01, 2010. Delo je prispelo 08. oktobra 2010, sprejeto 01. decembra 2010. 1 Univ. of Ljubljana, Biotechnical Fac., Dept. of Animal Science, Groblje 3, SI-1230 Domžale, Slovenia,, Ph.D., e-mail: gregor.gorjanc@bf.uni-lj.si 2 The same address as 1 3 Departamento de Mejora Genética, Instituto Nacional de Investigación Agraria, Carretera de La Coruña, km 7, 28040 Madrid, Spain, Ph.D. Simple reparameterization to improve convergence in linear mixed models Slow convergence and mixing are one of the main prob- lems of Markov chain Monte Carlo (McMC) algorithms applied to mixed models in animal breeding. Poor convergence is to a large extent caused by high posterior correlation between vari- ance components and solutions for the levels of associated ef- fects. A simple reparameterization of the conventional model for variance component estimation is presented which improves McMC sampling and provides the same posterior distributions as the conventional model. Reparameterization is based on the rescaling of hierarchical (random) effects in a model, which alleviates posterior correlation. The developed model is com- pared against the conventional model using several simulated data sets. Results show that presented reparameterization has better behaviour of associated sampling methods and is several times more efficient for the low values of heritability. Key words: statistics / mixed model / Bayesian analysis / McMC / reparameterization / convergence Enostavna reparametrizacija za izboljšanje konvergence line- arnih mešanih modelov Počasna konvergenca je eden največjih problemov upo- rabe metode Monte Carlo z Markovimi verigami (McMC) za mešane modele na področju genetike in selekcije domačih živali. Slaba konvergenca je v veliki meri posledica visoke posteriorne korelacije med komponentami variance in rešitvami za ravni pripadajočih vplivov. Predstavljamo enostavno reparametriza- cijo običajnega modela, ki izboljša lastnosti metode McMC in daje enake posteriorne porazdelitve parametrov modela kot standardni pristop. Reparametrizacija temelji na standardiza- ciji hierarhičnih (naključnih) vplivov v modelu, kar posledično spremeni posteriorne korelacije med parametri. Oba pristopa smo primerjali na večjem setu simuliranih podatkov. Rezultati kažejo, da reparametrizacija vodi do bolj učinkovitih metod McMC vzorčenja in je nekajkrat bolj učinkovita za analizo last- nosti z nizko heritabiliteto. Ključne besede: statistika / mešani model / bayesovska analiza / McMC / reparametrizacija / konvergenca 1 INTRODUCTION Mixed models are abundantly used in the field of animal breeding and genetics with the aim to infer genet- ic values of animals given some phenotypic and pedigree information (Henderson, 1984). In it simplest form the mixed model can be written as: y = Xb + Za + e, (1) where y is a vector of phenotypes, b is a vector of effects like sex, breed, age, etc., a is a vector of individual addi- tive genetic effects and e residual, ( ) ( )22 ,~| ee Np σσ I0e , while X and Z are design matrices linking effects to phe- notypic records. Pedigree information is included in the model hierarchically with prior distribution of individual additive genetic values, ( ) ( )22 ,~,| aa Np σσ A0Aa . Hend- erson (1972) developed the so called mixed model equa- tions (2) to efficiently obtain joint solutions for b and a, where 2aσAG = and 2 eσIR = : Acta agriculturae Slovenica, 96/2 – 201070 G. GORJANC et al.         =                + − − −−− −− yRZ yRX a b GZRZXRZ ZRXXRX 1 1 111 11 ˆ ˆ T T TT TT (2) Use of mixed model equations assumes known vari- ance components 2aσ and 2 eσ . Standard procedure is to estimate these variance components using restricted maximum likelihood method (REML; Patterson and Thompson, 1971) and to use these estimates in mixed model equations (2) ignoring the error of estimation in variance components. Another approach to statistical inference, Bayesian approach, treats inference of all model parameters joint- ly. Although conceptually very appealing, Bayesian ap- proach leads to formulas that are computationally intrac- table. This can be avoided by sampling methods such as Markov chain Monte Carlo (McMC; e.g., Gelman, et al., 2004). Wang et al. (1993) showed how McMC methods can be used with linear mixed models in animal breed- ing applications. In the case of linear mixed models all McMC computations follow from the posterior distribu- tion (3): (3) where prior distributions for and both variance compo- nents 2aσ and 2 eσ were assumed uniform (e.g., Gelman et al., 2004). Given that 2aσ and a are a priori correlated due to the prior definition of a, the a posteriori correla- tion between them is expected to be high. This leads to high autocorrelation between consecutive samples, mak- ing McMC method inefficient. Autocorrelations can be really problematic with low or near zero values for some variance components (e.g. additive genetic variance). This is caused by the shrinkage of a towards zero and in a next round of sampling variance component will again be close to zero, which can make the sampler stuck for quite some time at the values near zero (Gelman et al., 2004). Chib and Carlin (1999) proposed block sampling of some parameters in (2) to improve convergence. Autocorrelation has also been alleviated by the use of centered models (Gelfand et al., 1995), parameter ex- panded models (Liu and Wu, 1999; Gelman et al., 2003; Gelman, 2004) and data augmentation based models (Meng and van Dyk, 1997; van Dyk and Meng, 2001). These methods have been applied both to accelerate the Expectation-Maximization (EM) algorithm and to al- leviate the autocorrelation of McMC algorithms. In this work a reparameterization will be employed where addi- tive genetic values will be a priori uncorrelated with 2aσ . This approach will be compared against the conventional model of Wang et al. (1993). 2 METHOD Let us consider a simple animal model y = Xb + Za + e with the following distributional assumptions:            .,~| ,,~,| ,,~,,| 22 22 22 ee aa ee Np Np Np    I0e A0Aa IZaXbaby  (4) For this particular case and assuming uniform pri- ors for b and both variance components, ( ) .constp ∝b , ( ) .2 constp a ∝σ , and ( ) .2 constp e ∝σ , the equation (3) becomes:           ,exp exp|,,, 2 12 2 2 2 2 2 222               a T q e Tn a eeap     aAa ZaXbyZaXbyyab (5) where n is the number of records and q the number of animals. Full conditionals of the posterior  (5) can be sampled using the coefficient (left hand side) matrix of the mixed model equations (2), sums of squares, normal and scaled central χ−2 deviates (Wang et al., 1993). Here another approach is proposed, which allevi- ates the autocorrelation of samples from (5). It is based on the reparameterization of the model in the terms of a new augmented variable u, a = uσa. Such a model has been already proposed by Foulley and Quaas (1995) in a heterogeneous variance EM-REML context. To simplify the notation, σa is used instead of 2 aσ , but the model is still considered written in terms of 2aσ . The model is now y = Xb + Zuσa + e, with the following distributional assumptions:            .,~| ,,~| ,,~,,,| 22 222 ee eaea Np Np Np   I0e A0Au IZuXbuby  (6) The joint posterior distribution, assuming again uniform priors on b and both variance components, is:          ,exp exp|,,, 2 2 222 1 2 2 uAu ZuXbyZuXbyyub           T e a T a n eeap   (7) Note that in (7) variance component 2aσ drops out from the last part, but σa comes in the sum of squares of residuals. The full conditional distributions for the levels of both b and a are univariate normal distributions as in the conventional model, but considering a = uσa: ( ) ,,~,,,,| , 2 , ,,22       ∑ ∑≠ −− − ii e ii ij k kkijjii cc ucbcs eaii Nbp σσσ yub (8)         ,exp exp|,,, 1 2 1 1 2 122 2 1 2 1 aGaG ZaXbyRZaXbyRyab     T T eap  Acta agriculturae Slovenica, 96/2 – 2010 71 SIMPLE REPARAMETERIZATION TO IMPROVE CONVERGENCE IN LINEAR MIXED MODELS ( ) ,,~,,,,| , 2 , ,,22       ∑ ∑ ≠−− − ii e ii j ik kkijjii cc ucbcs eaii Nup σσσ ybu (9) where both si and ci,j are closely related with the conven- tional mixed model (2) but modified as:         =        + = − a T T ea T a T a TT σσσσ σ yZ yXS AZZXZ ZXXXC ,212 . (10) The full conditional distribution of 2eσ can be sam- pled from scaled inverted chi-square distribution with n − 2 degrees of freedom as in the conventional model:       2222 ~,,,|  naTaaep  ZuXbyZuXbyyub . (11) After some algebra the full conditional of 2aσ is:     , 2 2 exp,,,| 2 2 22              ZuZu ZuZu XbyZu yub TT e TT TT aa eap    (12) from which a truncated normal distribution can be rec- ognized when presented in terms of σa with mean   ZuZu XbyZu TT TT  , variance ZuZu TT e 2 , and truncation point at 0:     .0,,~,,,| 22        ZuZuZuZu XbyZuyub TT eTT TT TNp ea  (13) When the full conditional distribution of 2aσ does not involve the neighbourhood of zero, it is a scaled non- central χ2 distribution with 1 degree of freedom, with a scale parameter ZuZu TT e 2 and noncentrality parameter    .22 eTT TTT   ZuZu ZuXbyXbyZu     .,1~,,,| 222 2   ZuZu yub TT eeap (14) For cases where the posterior distribution of 2aσ is close to zero, the Metropolis-Hastings algorithm with positive proposal can be implemented, where the natural logarithm of the conditional density derived from (12) is:    ,,,,|ln 2222 2   aaeap yub (15) where τ represents mean and ρ variance from (13). 3 APPLICATION Seven simulated datasets were used to compare the length of burn-in period and Monte Carlo variance of the model y = Xb + Zuσs + e against the conventional sire model y = Xb + Zs + e. All datasets consisted of 10,000 records, 100 herds (b) and 500 unrelated sires (s). Records were randomly assigned to herds and sires, i.e., having on average 100 records per herd and 20 records per sire. True phenotypic variance was 100 and sire vari- ances for each simulated case were: 0.25, 0.5, 1.25, 2, 3.75, 5, and 10. Markov chain Monte Carlo method was imple- mented using Gibbs sampler for the full conditional dis- tributions described in (8, 9, and 11), while Metropolis sampler was used for sampling from (15). The length of burn-in period was determined by the use of coupling argument (Johnson, 1996; García-Cortés et al., 1998), where the tolerance of difference between two chains for the sire variance component was set to 10−4. After the burn-in period, chains with 20,000 samples were produced. Monte Carlo error was calculated empirically after 50 replicates for each simulated dataset. Presented True h2 Conventional model Reparametrized model 0.01 569.6 ± 266.1 9.8 ± 6.4 0.02 332.7 ± 165.2 8.4 ± 3.9 0.05 173.9 ± 37.1 7.8 ± 2.6 0.10 162.4 ± 41.2 7.8 ± 2.9 0.15 55.1 ± 5.8 6.8 ± 2.4 0.20 42.6 ± 2.7 7.4 ± 2.2 0.40 25.2 ± 3.6 8.4 ± 3.5 Table 1: Average (± standard deviation obtained empirically from 50 replicates) burn-in length by model and true heritability (h2) Preglednica 1: Povprečna (± standardni odklon, pridobljen empirično iz 50 ponovitev) dolžina ogrevalne faze glede na model in dejansko vrednost heritabilitete (h2) results show the rate of convergence in the terms of burn- in period (Table 1) and after burn-in period (Table 2) for the conventional model (4) and the new reparameterized model (6). Reparameterization of the model resulted in sub- stantial reduction in burn-in phase of McMC procedure (Table  1), especially with the low values of heritability. Inspection of trace plots (not shown) showed that in the case of low heritability values for additive genetic vari- ance were very close to zero as well as individual additive genetic values, which is expected. However, conventional model was prone to stuck in that configuration, while reparameterized model more easily explored wider pa- Acta agriculturae Slovenica, 96/2 – 201072 G. GORJANC et al. rameter space, which in turn leads to faster convergence to stationary distribution (e.g., Gelman et al., 2004). Both models gave the same posterior mean on aver- age (Table 2) for variance between sires. Only results for this effect are reported as this is one of the parameters that are hard to accurately estimate in linear mixed mod- els (e.g., Gelman et al., 2004). Posterior means for vari- ance between sires were larger than the true value. This can be attributed to skewed posterior distributions for this effect. Monte Carlo variance obtained after 50 rep- licates of conventional analysis was sensitive to the value of the true heritability, while this was not the case for reparameterized model. In addition, Monte Carlo vari- ance was higher with conventional model for heritabili- ties up to 0.1. More stable behaviour of reparameterized model was due to the possibility of easier escape from the neighbourhood of zero value for variance between sires. This means that reparameterized model is of a great value when traits with low heritability are analysed. 4 DISCUSSION The new data augmentation scheme resulted in an algorithm faster than the conventional Gibbs sampler for linear mixed models. Estimates for variance components do not suffer from getting stuck when visiting values close to zero and then the rate of convergence does not depend on the true value of heritability. When new model was applied to data sets with small heritability, Monte Carlo variance was around five times smaller. Therefore, the new model needs about twenty five times shorter chains to get the same Monte Carlo variance as the conventional model of Wang et al. (1993). The new model can be easily implemented in existing programs for the conventional model – slightly modifying the mixed model equations according to (10) and using the Metropolis algorithm to sample from the full conditional density of 2aσ . Our procedure is very similar to the parameter expanded models presented in (Liu and Wu, 1999; Gel- man et al., 2003; Gelman, 2004) among others for both the most frequent EM and Bayesian McMC. Their ap- proach also standardizes the additive genetic values, but in terms of a = uα, where α represents an extra augmented variables in the model, while our approach standardizes breeding values with its hyper-parameter, i.e., σa. The data augmentation scheme presented here can be understood as a particular case of that presented in van Dyk and Meng (2001), which is based on linear transformations of random variables, such as y = Xb + Zp+ e, where p = Yu + γ. In our case Y = Iσa− 1 and γ = 0, is the simplest case having a significant reduction of the Monte Carlo variance. Reparameterized model has been tested with a sire model example. Further research is necessary for ani- mal models or multiple trait models (Henderson, 1984), where the amount of missing information may be higher causing more stringency in standard McMC samplers. In such cases reparameterization in terms of u is expected to provide even better results than presented here. 5 CONCLUSION In summary, reparameterization of hierarchical ef- fects resulted in a feasible Markov chain Monte Carlo al- gorithm that accelerates the convergence of the conven- tional sampling methods for Bayesian analysis of linear mixed models. This procedure requires a little program- ming effort for implementation by researchers who have experience with the conventional sampling methods. 6 REFERENCES Chib S., Carlin B.P. 1999. On McMC sampling in hierarchical longitudinal models. Statistics and Computing, 9, 1: 17–26 van Dyk D.A., Meng X.L. 2001. The art of data augmentation (with discussion). Journal of Graphical and Computational Statistics, 10, 1: 1–111 Foulley J.L., Quaas R.L. 1995. Heterogeneous variances in Gaussian linear mixed models. Genetics, Selection and Evolution, 27,3: 211–228 García-Cortés L.A., Rico M., Groeneveld E. 1998. Using cou- pling with the Gibbs sampler to assess convergence in ani- mal models. Journal of Animal Science, 76, 2: 441–447 Gelfand A.E., Sahu S.K., Carlin B.P. 1995. Efficient parameteri- sations for normal linear mixed models. Biometrika, 82: 479–488 Gelman A. 2004. Parameterization and Bayesian model- True 2sσ True h2 Conventional model Reparametrized model 0.25 0.01 0.39 ± 0.03 0.38 ± 0.01 0.50 0.02 0.91 ± 0.03 0.98 ± 0.01 1.25 0.05 1.45 ± 0.02 1.44 ± 0.01 2.50 0.10 1.69 ± 0.02 1.69 ± 0.01 3.75 0.15 4.39 ± 0.01 4.39 ± 0.01 5.00 0.20 6.02 ± 0.01 6.03 ± 0.01 10.00 0.40 13.03 ± 0.01 13.05 ± 0.02 Table 2: Posterior mean (± standard deviation obtained empiri- cally from 50 replicates) for the component of variance between sires by model and true heritability (h2) Preglednica 2: Posteriorno povprečje (± standardni odklon, pri- dobljen empirično iz 50 ponovitev) komponente variance med očeti glede na model in dejansko vrednost heritabilitete (h2) Acta agriculturae Slovenica, 96/2 – 2010 73 SIMPLE REPARAMETERIZATION TO IMPROVE CONVERGENCE IN LINEAR MIXED MODELS ling. Journal of American Statistical Association, 99, 466: 537–545 Gelman A., Carlin J.B., Stern H.S., Rubin D.B. 2004. Bayesian data analysis. Chapman & Hall / CRC, 2 edition Gelman A., Huang Z., van Dyk D.A., Boscardin W.J. 2003. Transformed and parameter-expanded Gibbs samplers for multilevel linear and generalized linear model. Technical report, Departament of Statistics. Columbia University Henderson C.R. 1972. Sire evaluation and genetic trends. In: Proceedings of the Animal Breeding and Genetics Sympo- sium in Honor of Dr. J.L. Lush, Champaign, 29 jul. 1972. ASAS, ADSA, PSA: 10–41 Henderson C.R. 1984. Applications of Linear Models in Animal Breeding. Guelph, University of Guelph Johnson V.E. 1996. Studying convergence of Markov chain Monte Carlo algorithms using coupled sample paths. Jour- nal of American Statistical Association, 91, 433: 154–166 Liu J.S., Wu Y. 1999. Parameter expansion for data augmenta- tion. Journal of American Statistical Association, 94, 448: 1264–1274 Meng X.L., van Dyk D.A. 1997. The EM algorithm – an old folk-song sung to a fast new tune (with discussion). Journal of Royal Statistical Society, B Statistical Methodology, 59, 3: 511–567 Patterson H.D., Thompson R. 1971. Recovery of inter-block in- formation when block sizes are unequal. Biometrics, 58, 8: 545–554 Wang C.S., Rutledge J.J., Gianola D. 1993. Marginal inferences about variance components in a mixed linear model using Gibbs sampler. Genetics, Selection and Evolution, 25, 1: 41–62 Acta argiculturae Slovenica, 96/2, 75–80, Ljubljana 2010 doi:10.2478/v10014-010-0018-9 COBISS: 1.01 Agris category code: L01 MULTIVARIATE ANALYSIS OF PHENOTYPIC DIFFERENTIATION IN BUNAJI AND SOKOTO GUDALI CATTLE Abdulmojeed YAKUBU 1, 2, Kingsley Omogiade IDAHOR 1, Hadiza Salihu HARUNA 1, Matthew WHETO 3, Samuel AMUSAN 3 Received June 07, 2010; accepted September 06, 2010. Delo je prispelo 07. junija 2010, sprejeto 06. septembra 2010. 1 Nasarawa State Univ., Fac. of Agriculture, Dept. of Animal Science, Keffi, Shabu-Lafia Campus, P.M.B. 135, Lafia, Nigeria 2 Corresponding author’s e-mail: abdul_mojeedy@yahoo.com 3 Univ. of Agriculture, Dept. of Animal Breeding and Genetics, Abeokuta, Nigeria Multivariate analysis of phenotypic differentiation in Bunaji and Sokoto Gudali cattle The study aimed at examining morphometric differentia- tion in two Nigerian breeds of cattle using multifactorial dis- criminant analyses. Ten morphological traits (withers height, rump height,chest circumference, body length, face length, tail length, rump length, head width, rump width and shoulder width) of 224 Bunaji and 87 Sokoto Gudali cattle were mea- sured. The animals, which were aged 2.5−3.6 years, were sub- jected to extensive management system. The linear type traits of Sokoto Gudali cattle were significantly (P < 0.05) higher than those of their Bunaji counterparts, with the exception of body length and face length respectively. The stepwise discriminant analysis gave a better resolution as only three variables, rump width, withers height and face length were more discriminat- ing in separating the two cattle breeds. The Mahalanobis dis- tance (7.19) between the two cattle populations was high and significant, which is an indication that they belong to geneti- cally different groups. This was complemented by the result of the Nearest Neighbour Discriminant Analysis, where 85.48% of Bunaji cattle were classified into their source population while 96.55% of their Sokoto Gudali counterparts were correctly as- signed into their source genetic group. The present phenotypic information will be the basis for the establishment of further characterization, conservation and selection strategies for the two Nigerian breeds of cattle. Key words: cattle / breeds / morphological traits / dis- criminant analysis / characterization / Nigeria Multivariatna analiza fenotipskih razlik med “Bunaji” in “Sokoto Gudali” govedom V študiji smo z multivariatno diskriminantno analizo proučevali morfometrične razlike med dvema nigerijskima pasmama goveda. Merili smo deset morfoloških lastnosti (višina vihra, višina trupa, obseg prsi, dolžina telesa, dolžina glave, dolžina repa, dolžina trupa, širina glave, širina trupa in širina pleč) pri 224 živalih pasme “Bunaji” in 87 živalih pasme “Sokoto Gudali”. Živali so bile v ekstenzivni reji, stare med 2,5 ter 3,6 leti. Izmerjene vrednosti za linearne lastnosti živali pas- me “Sokoto Gudali” so bile statistično značilno večje (P < 0,05) kot pri živalih pasme “Bunaji”, izjema sta bila le dolžina telesa in dolžina glave. Za doseganje boljše resolucije smo uporabili postopno diskriminantno analizo, ker so le tri spremenljivke, širina telesa, višina vihra in dolžina glave, omogočile zanesljivo ločevanje obeh pasem. Mahalanobijeva distanca (7,19) med obema pasmama je bila visoko statistično značilna, kar nakazu- je, da populaciji pripadata različnim pasemskim skupinam. Te rezultate potrjuje tudi diskriminantna analiza najbližjih sose- dov, kjer je bilo 85,48% “Bunaji” goveda razvrščenega v izvorno populacijo, medtem, ko je bil ta odstotek pri “Sokoto Gudali” pasmi še višji (96,55). Tako pridobljene fenotipske informacije bomo uporabili za še natančnejši opis , zaščito in oblikovanje rejske strategije obeh nigerijskih pasem goveda. Ključne besede: govedo / pasme / morfološke lastnosti / diskriminantna analiza / karakterizacija / Nigerija Acta agriculturae Slovenica, 96/2 – 201076 A. YAKUBU et al. 1 INTRODUCTION The wide range of breeds and species that have evolved in various environments represent unique sets of genetic diversity. Genetic diversity has been defined as the variety of alleles and genotypes present in a popula- tion, and this is reflected in morphological, physiologi- cal and behavioural differences between individuals and populations (Frankham et al., 2002). It is generally ac- cepted that the highest amount of genetic diversity in the populations of livestock is found in the developing world where record keeping is poor, and the risk of extinction high and on the increase. Recently, loss of genetic diver- sity within indigenous livestock breeds has been a major concern (Kastelic et al., 2005). Every year many species and breeds of animals become extinct thereby decreas- ing the biodiversity and genetic variation of populations. Thus, breeds and species that have a tradition of breeding for many a centuries, a unique genotype and aesthetic and cultural value are being lost (Macijauskiene and Juras, 2003; Adamczyk et al., 2008). Hence, need for sustainable management and conservation strategies for these ani- mal genetic resources. Since the breed is the operational unit for the assessment of livestock diversity all over the world (Duchev and Groeneveld, 2006), contributions to characterization of local domestic animal populations are of major importance in developing countries. Characterization of livestock breeds is the first ap- proach to a sustainable use of its animal genetic resources (Lanari et al., 2003). The first step of the characterization of local genetic resources is based on the knowledge of variation in the morphological traits (Delgado et al., 2001). Morphometric measurements have been used to evaluate the characteristics of various breeds of animals, and could provide useful information on the suitability of animals for selection (Nesamvuni et al., 2000; Rastija et al., 2004; Araujo et al., 2006; Mwacharo et al., 2006; Martins et al., 2009; Yakubu, 2010). The outcome of ge- netic improvement programmes could also be evaluated on morphological basis (Riva et al., 2004). Although recent analyses have focused on molecular techniques, most mammalian species and subspecies originally were described on the basis of morphological characteristics (Feldhamer et al., 2004). Previous efforts on the pheno- typic characterization of breeds of livestock have been restricted to the use of analysis of variance, whereas the current trend in livestock classification involves the use of multivariate statistical tools (Traore et al., 2008; Yaku- bu and Akinyemi, 2010). This is because univariate sta- tistical analysis, according to Dossa et al. (2007), analyze each variable separately and do not explain how the pop- ulations under investigations differ when all measured morphological variables are considered jointly. Multi- factorial discriminant analyses have been found to be more suitable in assessing variation within a population and can discriminate different population types when all measured morphological variables are considered jointly. Cattle are the single most important livestock spe- cies in Nigeria in terms of animal protein, value and bio- mass (Tewe, 1998). However, information is scanty on the morphological characteristics of indigenous cattle especially the Bunaji and Sokoto Gudali which constitute 37.2 and 31.6% of the Nigerian cattle herd of 13,770,641 (RIM, 1992). The research questions are: How morpho- logically heterogeneous are Nigerian breeds of cattle? And has the classification of Nigerian cattle into different breeds any scientific support? The general objective of the study is to characterize two indigenous cattle breeds of Nigeria based on morphological variation using multi- variate discriminant analyses, which could help in proper management, conservation and genetic improvement of the local stock. 2 MATERIALS AND METHODS 2.1 EXPERIMENTAL ANIMALS AND LOCATION OF STUDY The experiment made use of a random sample of 211cattle of both sexes, comprising 124 Bunaji and 87 Sokoto Gudali, respectively. The animals were 2.5−3.6 years old as determined by dentition. They were reared through the extensive management system and originat- ed from different herds sampled in Nasarawa state, north central Nigeria. Efforts were made to restrict sampling to phenotipically pure Bunaji and Sokoto Gudali cattle respectively by measuring only those that conformed to the classification descriptors of both breeds. 2.2 MEASURED TRAITS Ten morphometric traits were measured on each animal. The body parameters were withers height (WH), rump height (RH), chest circumference (CC), body length (BL), face length (FL), tail length, rump length (RL), head width (HW), rump width (RW) and shoulder width (SW). Anatomical reference points were as earlier described (Yakubu et al., 2009). The height measurement (cm) was done using a graduated measuring stick. To achieve this, animals were placed on a flat ground and held by two field assistants. The length and circumfer- ence measurements (cm) were effected using a tape rule while the width measurements (cm) were taken using a calibrated wooden calliper. All measurements were car- Acta agriculturae Slovenica, 96/2 – 2010 77 MULTIVARIATE ANALYSIS OF PHENOTYPIC DIFFERENTIATION IN BUNAJI AND SOKOTO GUDALI CATTLE ried out by the same person in order to avoid between- individual variations. 2.3 STATISTICAL ANALYSIS The morphological traits were subjected to analysis of variance to determine genotype effect using the MEAN procedure of SPSS (2001). Means were separated using the two-tailed, two-sample t-test of the same statistical package. Stepwise discriminant procedure (SAS, 1999) was applied using PROC STEPDISC to determine which morphological traits have more discriminant power than others. The relative importance of the morphometric variables in discriminating between the two cattle popu- lations was assessed using the level of significance, partial R2 and F-statistic. The CANDISC procedure was used to perform univariate and multivariate one-way analysis that calculated the Mahalanobis distance between the two cattle breeds. The ability of these canonical functions to assign each individual animal to its breed was calcu- lated as the percentage of correct assignment to each genetic group using the DISCRIM procedure (Nearest Neighbour Discriminant Analysis). 3 RESULTS AND DISCUSSION Descriptive statistics of the morphological traits of Bunaji and Sokoto Gudali cattle are presented in Table 1. Generally, the linear body measurements of Sokoto Gu- dali were significantly (P < 0.05) higher than those of the Bunaji cattle with the exception of body length and face length respectively. Comparative measurements of morphometric traits can provide evidence of breed re- lationships and size. The considerable variation in body dimensions of the two cattle breeds might not be uncon- nected with individual breed’s potential and peculiari- ties. While the Bunaji cattle is noted for milk production, their Sokoto Gudali counterparts which rank second in milk production produce more meat and appear to have more draught power than the former. The estimates ob- tained for height at withers of adult cattle in this study are comparable to those of the Nandi (110–122 cm), Mongalla (100–110 cm) (Rege, 1999), Mexican Criollo Chinampo (101–117 cm) (Espinoza et al., 2009) and Su- dan Baggara (115.9–148.80 cm) (Alsiddig et al., 2010) cattle, respectively. The chest circumference values are, however, higher than the range of 122–127 cm reported for North Bengal Grey cattle in Bangladesh (Al-Amin et al., 2007). Traits Bunaji Sokoto Gudali Mean ± SE SD CV Mean ± SE SD CV Withers height Rump height Chest circumference Body length Face length Tail length Rump length Head width Rump width Shoulder width 111.84 ± 0.98b 120.34 ± 1.01b 141.94 ± 1.62b 175.29 ± 2.25a 52.88 ± 0.49a 76.81 ± 0.97b 39.06 ± 0.42b 15.54 ± 0.14b 33.32 ± 0.44b 28.94 ± 0.43b 10.87 11.20 18.07 25.04 5.48 10.80 4.73 1.60 4.95 4.77 9.72 9.31 12.73 14.28 10.36 14.06 12.11 10.30 14.86 16.48 127.50 ± 0.53a 149.53 ± 1.55a 181.15 ± 1.92a 179.02 ± 1.55a 53.28 ± 0.34a 84.27 ± 0.41a 42.17 ± 0.31a 21.15 ± 0.41a 50.43 ± 1.02a 31.79 ± 0.28a 4.97 14.43 17.89 14.41 3.19 3.87 2.91 3.80 9.47 2.58 3.90 9.65 9.88 8.05 5.99 4.59 6.90 17.97 18.78 8.12 Table 1: Descriptive statistics of morphological traits of Bunaji and Sokoto Gudali cattle Preglednica 1: Opisna statistika morfoloških lastnosti “Bunaji” in “Sokoto Gudali” goveda SE – Standard error, SD – Standard deviation, CV – Coefficient of variation. Means in the same row with different superscripts are significantly different (P < 0.05) Step Variables entered Partial R2 F-value Pr > F Wilk’s Lambda Pr < lambda Average squared canonical correlation Pr > ASCC 1 RW 0.5824 291.54 < 0.0001 0.417550 < 0.0001 0.582 < 0.0001 2 WH 0.0948 21.67 < 0.0001 0.362555 < 0.0001 0.637 < 0.0001 3 FL 0.0408 8.85 0.0033 0.400504 < 0.0001 0.599 < 0.0001 Table 2: Summary of stepwise selection of traits Preglednica 2: Povzetek postopnega izbora lastnosti RW – rump width, WH – withers height, FL – face length. Acta agriculturae Slovenica, 96/2 – 201078 A. YAKUBU et al. The stepwise discriminant analysis showed that rump width, withers height and face length were the most discriminating variables between Bunaji and Sokoto Gu- dali cattle (Table 2). Their respective partial R2 and F- values were 0.5824, 0.0948 and 0.0408; 291.54, 21.67 and 8.85 with high significant values (P < 0.01–P < 0.0001). Morphological variables are easy to monitor and may fa- cilitate the use of ethnological characterization and at the same time institute reliable racial discriminants (Herrera et al., 1996). The three morphological variables obtained in the present study are more important and informative, and could be used to assign the two cattle breeds into previous workers on goats (Dossa et al., 2007, Yakubu et al., 2010a,b and c), sheep (Traore et al. 2008; Yakubu and Akinyemi, 2010), cattle (Ndumu et al., 2008) and buffalo (Johari et al., 2009) respectively. The general aim of genetic conservation is to main- tain within and across breed diversity, where within breed diversity refers to the genetic management of one population and the across breed diversity implies the ge- netic management of many populations. Within breed diversity it is needed for the breed to genetically adapt to changes in the production and economic environ- ment, and to avoid inbreeding problems. Across breed diversity is needed to provide alternatives if a breed hap- pens to run into genetic problems due to genetic drift or changes in the production systems (Meuwissen, 2009). Population studies which elucidate the relationship exist- ing between the different breeds of a given species may offer useful information for the conservation and man- agement of animal genetic resources (AnGR) such as the evolution of the breeds, the development of gene pools and the magnitude of genetic differentiation. According to Mariante et al. (2008), national AnGR conservation programmes should use the association of phenotypic data, molecular polymorphisms and adequate statistical methods which reflect the real condition of a population. This was buttressed by Berthouly et al. (2010) who stud- ied genetic diversity of Vietnamese H’mong cattle using multivariate analysis on morphometric and genetic data. The present information on the phenotypic differen- tiation of Bunaji and Sokoto Gudali could therefore be exploited in designing appropriate strategies for their management and conservation. However, there is a need for a genetic study using protein and DNA microsatellite markers to complement the results arisen from morpho- metric differentiation of the two most populous Nigerian breeds of cattle. 4 CONCLUSIONS This study showed that Sokoto Gudali had higher mean values in withers height, rump height, chest cir- cumference, tail length, rump length, head width, rump width and shoulder width compared to their Bunaji counterparts. The two cattle breeds were not significant- ly different in body length and face length respectively. However, rump width, withers height and face length were found to be the most discriminating variables to assign Bunaji and Sokoto Gudali cattle into distinct ge- netic groups. However, the present information on the morphometric differentiation of Bunaji and Sokoto Gu- dali breeds of cattle could be complemented with genetic characterization using biochemical and DNA markers. Breed Bunaji Sokoto Gudali Bunaji Sokoto Gudali 0 7.19 7.19 0 Table 3: Mahalanobis distance between Bunaji and Sokoto Gudali cattle Preglednica 3: Mahalanobijeve distance med “Bunaji” in “So- koto Gudali” govedom distinct populations, thereby reducing the errors of selec- tion in future breeding and selection programmes. The Mahalanobis distance matrix is given in Table 3. The pairwise distance (7.19) between the two cattle breeds was highly significant (P < 0.001). This was sub- stantiated by the classification result (posterior probabil- ity of membership in each population). While 85.48% of Bunaji cattle were classified into their source population, 96.55% of their Sokoto Gudali counterparts were cor- rectly assigned into their source genetic group (Table 4). The high morphological distance between the two cat- tle populations coupled with high correct assignment to source genetic groups is an indication that they belong to different breeds. This could have been facilitated by the fact that measurements were restricted to phenotypi- cally pure animals. The use of multivariate discriminant analyses therefore could be successfully used in morpho- metric differentiation. This is similar to the reports of Breed Bunaji Sokoto Gudali Bunaji Sokoto Gudali Error level Priors 85.48 3.45 0.15 0.50 14.52 96.55 0.03 0.50 Table 4: Percent (%) of individual cattle classified into breed Preglednica 4: Odstotek (%) osebkov, razvrščenih v posamezno pasmo Acta agriculturae Slovenica, 96/2 – 2010 79 MULTIVARIATE ANALYSIS OF PHENOTYPIC DIFFERENTIATION IN BUNAJI AND SOKOTO GUDALI CATTLE This could aid field assessment, management and con- servation of the two cattle populations, where the goal is to obtain phenotypically pure local genetic resources for future selection and breeding improvement strategies. 5 REFERENCES Adamczyk K., Felenczak A., Jamrozy J., Szarek J., Bulla, J. 2008. Conservation of Polish Red cattle. Slovak J. Anim. Sci., 41: 72–76 Al-Amin M., Nahar A., Bhuiyan A.K.F.H., Faruque M.O. 2007. On-farm characterization and present status of North Ben- gal Grey (NBG) cattle in Bangladesh. AGRI, 40: 55–64 Alsiddig M.A., Babiker S.A. Galal M.Y., Mohammed A.M. 2010. Phenotypic characterization of Sudan Zebu cattle (Baggara type). Res. J. Anim. Vet. Sci., 5: 10–17 Araujo J.P., Machado H., Cantalapiedra J., Iglesias A., Petim- Batista F., Colaco J., Sanchez L. 2006. Biometric analysis of Portuguese Minhota cattle. Proceedings 8th World Congress on Genetics Applied to Livestock Production, 13–18 Aug. 2006, Belo Horizonte, MG, Brazil Berthouly C., Maillard J.C., Phan Doan L., Nhu Van T., Bed’Hom B., Leroy G., Hoang Thanh H., Laloe D., Bruneau N., Vu Chi C., Nguyen Dang V., Verrier E., Rognon X. 2010. Revealing fine scale subpopulation structure in the Viet- namese H’mong cattle breed for conservation purposes. BMC Genetics, 11: 45, doi:10.1186/1471-2156-11-45 Delgado J.V., Barba C., Camacho M.E., Sereno F.T.P.S., Mar- tinez A. Vega-Pla J.L. 2001. Livestock characterisation in Spain. Animal Genetic Resources Information, 29: 7–18 Dossa L.H., Wollny C., Gauly M. 2007. Spatial variation in goat populations from Benin as revealed by multivariate analysis of morphological traits. Small Rumin. Res., 73: 150–159 Duchev Z., Groeneveld E. 2006. Improving the monitoring of animal genetic resources on national and international level. Arch. Tierz. Dummerstorf, 49: 532–544 Espinoza J.L., Guevara J.A., Palacios, A. 2009. Morphomet- ric and phaneroptic characterization of Mexican Criollo Chinampo cattle. Arch. Zootec., 58: 277–279 Feldhamer G.A., Drickamer L.C., Vessey S.H., Merritt J.F. 2004. Mammalogy: adaptation, diversity, ecology. McGraw Hill, Boston, Massachusetts, USA Frankham R., Ballou J.D., Briscoe D.A. 2002. Introduction to conservative genetics. Cambridge University Press. Herrera M., Rodero E., Gutierrez M.J., Pena, F., Rodero J.M. 1996. Application of multifactorial discriminant analysis in the morphostructural differentiation of Andalusian caprine breeds. Small Rumin. Res., 22: 39–47 Johari S., Kurnianto E., Sutopo S., Hamayanti W.A. 2009. Mul- tivariate analysis on phenotypic traits of body measurement in swamp buffalo (Bubalus bubalis). J. Indone. Trop. Anim. Agric., 34: 289–294 Kastelic M., Zan Lotric M., Kompan D. 2005. Linear body mea- surements of Cika cattle in comparison to Pinzgauer cattle. Acta Agriculturae Slovenica, 86, 2: 85–91 Lanari M.R., Taddeo H., Domingo E., Centeno M.P., Gallo L. 2003. Phenotypic differentiation of exterior traits in local Criollo goat population in Patagonia (Argentina). Arch. Tierz. Dummerstorf, 46: 347–356 Macijauskiene V., Juras R. 2003. An Attempt at analyzing the selected traits of body conformation, growth, performance and genetic structure of Lithuanian native Zemaitukai horse, the breed being preserved from extinction. Animal Science Papers and Reports, 21: 35–46 Mariante A.S., Egito A.A., Albuquerque M.S.M., Paiva S.R., Ramos A.S. 2008. Managing genetic diversity and society needs. Revista Brasileira de Zootecnia, 37, doi:10.1590/ S1516-35982008001300016 Martins C.E.N., Quadros S.A.F., Trindade J.P.P., Quadros F.L.F., Costa J.H.C., Raduenz G. 2009. Shape and function in Bra- ford cows: The body shape as an indicative of performance and temperament. Arch. Zootec., 58: 425–433 Meuwissen A., 2009. Genetic management of small popula- tions: A review. Acta Agriculturae Scand, Section A, 59: 71–79 Mwacharo J.M., Okeyo A.M., Kamande G.K. Rege J.E.O. 2006. The small East African shorthorn zebu cows in Kenya. 1: Linear body measurements. Trop. Anim. Health Prod., 38: 65–76 Ndumu D.B., Baumung R., Hanotte O., Wurzinger M., Okeyo M.A., Jianlin H., Kibogo H., Solkner J. 2008. Genetic and morphological characterization of the Ankole Longhorn cattle in the African Great Lakes region. Genet. Sel. Evol., 40: 467–490 Nesamvuni A.E., Mulaudzi J., Ramanyimi N.D., Taylor G.J. 2000. Estimation of body weight in Nguni-type cattle un- der communal management conditions. S. J. Anim. Sci., 30 (Supplement 1): 97–98 Rastija T., Baban M., Antunovic Z., Mandic I. 2004. A compari- son and development of morphometric characteristics of stallions and mares on the Lipizzaner stud of Dakovo. Acta Agriculturae Slovenica, Supplement 1: 195–200 Rege J.E.O. 1999. The state of African cattle genetic resources I. Classification framework and identification of threatened and extinct breeds. FAO/UNEP Animal Genetic Resources Information Bulletin, 26: 1–26 RIM 1992. Nigerian Livestock Resources: National Synthesis. Jersey, UK, Resource Inventory and Management Ltd. SAS. 1999. Statistical Analysis System User’s guide: Statistic. Cary, NC 27513, U.S.A., SAS Institute Inc. SPSS. 2001. Statistical Package for the Social Sciences. New York, SPSS Inc. Tewe O.O. 1998. Sustainability and Development: Paradigms from Nigeria Livestock Industry. University Inaugural Lec- ture delivered on October 9, 1997 on behalf of the Faculty of Agriculture and Forestry, University of Ibadan, Ibadan Yakubu A., Ogah D.M., Idahor K.O. 2009. Principal component analysis of the morphostructural indices of White Fulani cattle. Trakia J. Sci., 7: 67–73 Yakubu A. 2010. Path coefficient and path analysis of body weight and biometric traits in Yankasa lambs. Slovak J. Anim. Sci., 43: 17–25 Yakubu A., Salako A.E., Imumorin, I.G. 2010a. Multivariate analysis of spatial patterns of morphological traits in West African Dwarf goats in three agro-ecological zones of Nige- ria. J. Appl. Anim. Res., (in press) Acta agriculturae Slovenica, 96/2 – 201080 A. YAKUBU et al. Yakubu A., Salako A.E., Imumorin I.G., Ige A.O., Akinyemi M.O. 2010b. Discriminant analysis of morphometric differ- entiation in the West African Dwarf and Red Sokoto goats. S. Afr. J. Anim. Sci., 40, (in press) Yakubu A., Salako A.E., Imumorin I.G. 2010c. Comparative multivariate analysis of biometric traits of West African Dwarf and Red Sokoto goats. Trop. Anim. Health Prod. (ac- cept with minor revision) Yakubu, A. and Akinyemi, M.O. 2010. An evaluation of sexual size dimorphism in Uda sheep using multifactorial dis- criminant analysis. Acta Agriculturae Scand Section A, 60: 74–78 Acta argiculturae Slovenica, 96/2, 81–86, Ljubljana 2010 doi:10.2478/v10014-010-0019-8 COBISS: 1.01 Agris category code: P30, P34 ANTIBIOTIC-RESISTANT SOIL BACTERIA IN HIGH-ALTITUDE (5000–6000 m) SOILS OF THE HIMALAYA Blaž STRES 1 Received August 16, 2010; accepted December 03, 2010. Delo je prispelo 16. avgusta 2010, sprejeto 03. decembra 2010. 1 Univ. of Ljubljana, Biotechnical Fac., Dept. of Animal Science, Groblje 3, SI-1234 Domžale, Slovenia, Ph.D., E-mail: Blaz.Stres@bf.uni-lj.si, Fax: +386 1 72 41 005 Antibiotic-resistant soil bacteria in high-altitude (5000– 6000 m) soils of the Himalaya In this study, low-carbon soils collected from an altitude transect from 5000 m to 6000 m were adopted as a simple model system with lower interaction complexity. This could help disentangle the basic environmental factors shaping the abundance and distribution of expressed resistance traits in culturable portion of fast growing heterotrophic strains. Im- proved plate counts were performed at 4 °C using 0.01 diluted nutrient broth supplemented with cold soil extract as a general media and additionally supplemented with antibiotics Ampi- cillin, Erythromycin, Kanamycin and Tetracyclin. A number of colonies (500) isolated from six locations were also tested separately for their antibiotic resistance. The results show that these high-altitude cold soils contained bacterial populations culturable at 4 °C in the range of 106 cells / g that were resistant to the four antibiotics and their various combinations tested in this study. The highest prevalence of resistance was observed in vegetated soils, whereas almost two orders of magnitude lower abundance of resistant cells was cultured from barren soils. Re- dundancy analysis showed that vegetation, soil carbon and pH were successful in explaining the interaction between environ- mental parameters and various culturable fractions of cold soil bacteria used in this study. Key words: microbiology / bacteria / antibiotics / resis- tance / high-altitude / soil / interaction model Na antibiotike odporne bakterije v visokogorskih tleh Himalaje (5000–6000 m) V študiji sem uporabil vzorce tal z nizko vsebnostjo or- ganskega ogljika iz višinskega transekta 5000 m–6000 m kot po- enostavljen modelni sistem z nizko kompleksnostjo interakcij. Ta bi lahko pomagal razumeti osnovne okoljske dejavnike, ki uravnavajo porazdelitev in obseg izraženih rezistenčnih lastno- sti gojljivega dela hitro rastočih heterotrofnih sevov. Izboljšano štetje na ploščah sem izvedel pri 4 °C na 0,01 koncentriranem hranilnem bujonu, dopolnjenim s hladnim ekstraktom tal, kot splošnim gojiščem, ki sem ga dopolnil z posameznimi antibi- otiki (ampicilin, eritromicin, kanamicin in tetraciklin). Večje število izolatov (500) iz šestih lokacij sem prav tako testiral lo- čeno na njihovo odpornost na antibiotike. Ugotavljal sem tudi povezavo med okoljskimi dejavniki ter porazdelitvijo odpornih sevov in splošnega gojljivega deleža talnih bakterij. Rezultati kažejo, da visokogorska hladna tla vsebujejo pri nizkih tempe- raturah gojljive bakterijske populacije (106 / g), ki so odporni na posamezne antibiotike in razne njihove kombinacije, upo- rabljene v tej študiji. Poraščena tla imajo največji delež odpor- nih bakterij, skoraj dva reda manjši pa je prisoten v golih tleh. Statistična analiza je pokazala, da vegetacija, organski ogljik ter pH uspešno razložijo interakcijo med okoljskimi dejavniki in posameznimi gojenimi deleži bakterij, izoliranih iz hladnih tal. Ključne besede: mikrobiologija / bakterije / antibiotiki / rezistenca / visokogorje / tla / model interakcij Acta agriculturae Slovenica, 96/2 – 201082 B. STRES 1 INTRODUCTION The influence of the wide spread and long term use of antibiotics on the prevalence of resistance traits in the environment is still poorly understood although antibiotic resistance has been recognized as a global pub- lic health concern (for review see e.g. Allen et al., 2010; Nwosu, 2001). In this respect, understanding the role of environmental bacteria as resistance gene reservoir is one of the key problems (Demaneche et al., 2008; Riesen- feld et al., 2004; Nwosu, 2001). Many studies have shown the existence of a considerable pool of resistance genes in agricultural soils, fish farm sludges and waters, in iso- lated human populations even in the absence of an obvi- ous selection pressure (Allen et al., 2010). In addition, metagenomic studies have identified novel resistance genes, a much wider diversity of known genes belonging to various resistance gene families and novel genes cod- ing molecules and enzymes involved as potentiators of microbial resistance (Demaneche et al., 2008; Riesenfield et al., 2004; Allen et al., 2010). A simple model system composed of low-carbon soil altitude transect from 5000 m to 6000 m was adopt- ed. Low plant diversity and short growing season in one of the most remote and human least directly impacted regions served as a system with lower interaction com- plexity that could help disentangle the basic environ- mental factors shaping the distribution and abundance of expressed resistance traits in culturable portion of fast growing heterotrophic strains. Improved plate counts for this environment were performed at 4 °C using 0.01 di- luted nutrient broth supplemented with cold soil extract as a general media. In addition, a number of colonies (500) isolated from six locations were tested for their an- tibiotic resistance. The distribution of antibiotic strains was correlated to environmental parameters recorded and described before (Stres et al., 2010). 2 MATERIALS AND METHODS 2.1 GENERAL CULTURABILITY The soils and physical-chemical and various biolog- ical characteristics of the six soil samples were described before (Stres et al., 2010). Shortly, soils were collected on the south facing slope of high alpine ridge descending from Drohmo peak (6980 m), the Kanchenjunga Himal, Nepal. The abundance of the culturable fraction of heter- otrophic microbial community was assessed using plate counts according to approach described by Hashimoto and Hattori (1989) and Janssen et al. (2002) and modified as described below. The soil dilution series was prepared in 1 g / L MgSO4 buffer and three replicates per dilution were plated on the following three different oligotrophic complex media for each sample: 0.01 strength Nutrient Broth (Difco) in salt solution solidified with 1% agar (NB-A) supplemented with cold soil extract (Ley et al., 2001; Janssen et al., 2002; Olsen and Bakken, 1987). A colony – forming curve (CFC) (Hashimoto and Hattori, 1989) was generated for each soil by counting newly vis- ible colonies over a 14 week incubation period at 4 ºC and plotting the cumulative number of colonies at each time point. Only the counts after three weeks were used for calculations in the present work. 2.2 GENERAL RESISTANCE TO ANTIBIOTIC COMPOUNDS. The 10 g portions of soil samples were resuspended in total volume of 100 mL of sterile salt solution (Ley et al., 2001) and the cells were stripped from soil particles at 200 rpm for 20 min. Decimal serial dilutions were pre- pared and 100 μL were inoculated on NB-A plates sup- plemented with one of the four antibiotic compounds (Ampicillin (50 µg / mL), Tetracycline (20 µg / mL), Kanamycin (20 µg / mL), Erythromicin(15 µg / mL) and incubated at 4 ºC. Plates were inspected for well – spaced colonies (distance > 5 mm) after 3 weeks as no additional colonies appeared after 4 week incubation and 95% con- fidence intervals were calculated. To verify the antibiotic resistance of the strains ap- pearing after longer incubation periods, a subset of ran- domly selected colonies (n = 50) from each sample was restreaked on the same plates supplemented with single antibiotic compound. 2.3 ANTIBIOTIC RESISTANCE OF ISOLATED STRAINS In order to obtain a more conservative estimate on resistant fraction within the culturable portion of bacte- ria, strains obtained from the first cultivation experiment without antibiotic compounds were tested separately for their antibiotic resistance. Cultures were plated on the same media they were isolated on, but supplemented with one of the four antibiotics as above. 2.4 STATISTICAL ANALYSES The antibiotic resistance of isolated strains obtained in this study and environmental parameters (Stres et al., 2010) served as input data in linear constrained ordina- Acta agriculturae Slovenica, 96/2 – 2010 83 ANTIBIOTIC-RESISTANT SOIL BACTERIA IN HIGH-ALTITUDE (5000–6000 m) SOILS OF THE HIMALAYA tion, redundancy analysis (RDA) with forward selection that was used to create an environmental model explain- ing the variability in response variables (antibiotic resist- ance patterns, abundance of resistant colonies, general abundance of culturable cells). The Monte Carlo permu- tation test (999 permutations) was applied to compute the significance of hypothetical relations using CANO- CO V 4.5 (Biometris) (Leps and Smilauer, 2002). 3 RESULTS AND DISCUSSION The abundance of resistant CFU to four antibiotics used ranged from lowest 102 to 106 CFU / g soil in barren and plant covered soils, respectively. Antibiotic resistant CFU determined at 4 °C were almost 100 times more abundant in plant covered (5200 m, 5400 m, 5600 m) than in barren soils (5000 m, 5800 m, 6000 m), despite rather similar number of culturable cells in these soils (Fig. 1). There was no discernible effect of particular antibi- otic compound on the abundance of resistant CFU with- in particular soil sample, however, the levels of resistant colonies were significantly different (P < 0.01) between barren and vegetated soils. The percentages of resistant bacteria varied from 0.01 to 15% in barren soils, median 2%. Surprisingly, the number of antibiotic resistant and the number of culturable bacteria appeared to be equal in plant covered soils, suggesting that all culturable bacte- ria were also resistant to antibiotics. This is surprising as the values reported in this study are one to two orders of magnitude higher than those reported for transgenic and control corn fields for Ampicillin resistance. In addition, the prevalence of Ampicillin resistant bacteria in un- disturbed prairie soil ranged from 54.4% to 69.9% (De- maneche et al., 2008), representing half of the prevalence found in this study. The results of the two studies could represent a simple gradient from intensive agricultural practice through undisturbed prairie to simplified more extreme natural vegetated environment where antibiotic resistance could represent a novel competitive advantage. However, whether these strains are more exposed to an- tibiotic producing strains or are only exposed to better conditions for gene exchange can not be resolved. Seem- ingly the question, whether the antibiotics used in this study serve as activators of specific biochemical path- 0 1 2 3 4 5 6 7 8 9 6000m 5800m 5600m 5400m 5200m 5000m Samples Lo g CF U Em Cn Amp Tc Culturable Figure 1: Colony counts of heterotrophic and antibiotic resistant strains emerging on 0.01 NB-A plates supplemented with single antibiotic compound at 4 °C. Error bars represent 95% confidence intervals. Amp – Ampicillin; Em – Erythromycin; Kn – Kanamycin; Tc – Tetracycline. Slika 1: Število heterotrofnih in na antibiotike odpornih konij, ki so zrasle na gojišču 0,01 NB-A s posameznimi antibiotiki pri 4 °C. Oznake napak predstavljajo 95 % intervale zaupanja. Amp – ampicilin; Em – erithromicin; Kn – kanamicin; Tc – tetraciklin. Acta agriculturae Slovenica, 96/2 – 201084 B. STRES ways, signaling molecules in quorum sensing or just as a simple carbon source can not be answered at this time. The technical limitations and differences in ap- proaches could be also limiting the comparability of the results between studies, as different approaches to cell stripping, temperature and time of incubation were used next to different carbon source. This highlights the pro- found inconsistencies in the approaches used to moni- tor the resistance properties of environmental bacteria as these approaches are not standardized and the data are produced on a range of media, antibiotic concentrations, temperatures and incubation periods (Nwosu, 2001; D’Costa et al., 2006; Allen et al., 2010). However, it is tempting to speculate that these re- sults indicate that antibiotic resistance is a common trait in this high altitude environment and that plant pres- ence significantly increased the frequency of antibiotic resistance to one and combinations of multiple antibiot- ics. It also seems that these resistance traits are acquired through different mechanisms than human application and indicate that cold soil bacteria are an important res- ervoir of antibiotic resistance genes potentially entering water flows during enhanced percolation during snow thaw. Further, the testing of the strains isolated on NB- A-CSA plates without antibiotic compounds revealed that the vast majority of resistant strains were resistant to three antibiotics, Ampicillin, Kanamycin and Eryth- romycin (Fig. 2). This is congruent with the recent find- ings of D’Costa et al. (2006) that environmental strains are resistant to multiple antibiotics and also suggests that the distribution of resistance determinants is rather simi- lar among the antibiotic resistance strains from the six samples of the high-altitude cold soils. In addition, the overall abundance of resistant population appears to be modulated exclusively by the presence of plants, despite the similar abundances of culturable bacteria in other barren samples, differences in soil chemistry and plant species (Poa sp., mosses or combination of the two) cov- ering vegetated soils (Stres et al., 2010). In order to establish which environmental param- eters were significantly associated with the observed pat- terns in antibiotic resistance patterns in the cold soils, RDA analysis was conducted. Environmental character- istics (soil physical and chemical parameters) reported in Figure 2: The distribution of antibiotic resistance determinants in analyzed strains isolated from six composite high-altitude cold-soil samples from 5000 – 6000 m altitude transect. The letters designate the antibiotics and their combinations. A- Ampicillin; E – Erythro- mycin; K – Kanamycin; T – Tetracycline. Slika 2: Porazdelitev determinant odpornosti analiziranih sevov izoliranih iz šestih visokogorskih hladnih tal iz transekta med 5000–6000 m nadmorske višine. Oznake napak predstavljajo antibiotike ali njihove kombinacije: A – ampicilin; E – erithromicin; K – kanamicin; T – tetraciklin. Acta agriculturae Slovenica, 96/2 – 2010 85 ANTIBIOTIC-RESISTANT SOIL BACTERIA IN HIGH-ALTITUDE (5000–6000 m) SOILS OF THE HIMALAYA Stres et al. (2010) served as explanatory variables where- as the general culturability and abundance of resistant CFU to each antibiotic (Fig. 1) served as response vari- ables. RDA showed that only three out of 20 measured environmental parameters could explain significantly the variability in the measured abundances of culturable cells and resistant populations. The best predictor of variabil- ity in the high-altitude microbial abundance at 4 °C was vegetation, organic carbon and pH, explaining 87.1%, 9.1% and 3% of the data variability (P = 0.002; P = 0.026; P = 0.01), respectively. This environmental model ex- plained 99.2% of variability in abundance of the various culturable fractions explored in this study and 99.7% of species – environment relations. Other environmental parameters tested in this study (Stres et al., 2010) did not produce significant effects and were omitted from Fig. 3 with two exceptions (MWI(DOC), sugars). Interestingly, the soil content of reductive sugars (Stres et al., 2010) was directly correlated to general cul- turability of soil bacteria, however, this correlation was not found statistically significant. This finding is interest- ing in its own right in understanding of environmental parameters that enable recovery of larger proportions of culturable bacteria from the environmental sam- ples (Janssen et al., 2002; Davies et al., 2005). This ap- proach could provide a different strategy in cultivation approaches, first by analyzing the environmental condi- tions in various samples and pinpointing the environ- mental parameters correlated to increased culturabiltiy of microorganisms, with efforts mostly directed to various organic species, which is now much more easily achiev- able through the use of GC-MS or MALDI-TOF MS. On the other hand, the molecular weight index describing the size of complex organic substances was inversely proportional to general culturability. This is also interest- ing as the size of this index is inversely proportional to molecular weight, suggesting that the general measure of an average molecular weight in dissolved organic carbon fraction offers a too low resolution to be of any particu- lar value in such cases. On the other hand, RDA showed no significant correlation between patterns of antibiotic resistance (Fig. 2) and environmental parameters. This suggests that the distribution of antibiotic resistance de- terminants did not differ significantly from a random pattern. Alternatively, the presence of other factors and processes that shape the distribution of particular anti- Figure 3: Results of redundancy analysis (RDA) describing general cultivation success and the abundance of various resistant popula- tions (the response variables) in relation to the sampling localities (empty circles) and environmental parameters (bold arrows). The significant environmental parameters are marked with asterisk (*). Corg – organic carbon; MWI(DOC) – molecular weight index of dissolved organic carbon; Amp – Ampicillin; Em – Erythromycin; Kn – Kanamycin; Tc – Tetracycline. Slika 3: Rezultati statistične analize (RDA) prikazujejo uspešnost gojenja heterotrofnih mikroorganizmov ter različnih odpornih populacij (odzivne spremenljivke – črtkane puščice) v odvisnosti od mest vzorčenja (krožci) ter okoljskih dejavnikov (poudarjene puščice). Signifikantni okoljski dejavniki so označeni z zvezdico (*). Corg – organski ogljik; MWI(DOC) – indeks velikosti molekul raztopljenega organskega ogljika; Amp – ampicilin; Em – erithromicin; Kn – kanamicin; Tc – tetraciklin. Acta agriculturae Slovenica, 96/2 – 201086 B. STRES biotic patterns, not recorded in this study, could explain these observations. 4 CONCLUSIONS The high-altitude cold-soils contain at 4 °C cultur- able bacterial populations that are resistant to the four antibiotics tested in this study. The highest prevalence of resistance to antibiotics was recorded for plant covered soils, where all culturable cells exhibited resistance to antibiotics. On the contrary, almost two orders of mag- nitude lower abundance of resistant cells was cultured in barren soils. Redundancy analysis showed that veg- etation, soil carbon and pH were successful in explain- ing the interaction between environmental parameters and culturable fractions of cold soil bacteria used in this study. 5 ACKNOWLEDGMENTS I am indebted to Dr. Jerneja Ambrožič Avguštin for the stimulating discussions during our sabbatical at the Mediterranean Institute for Advanced Studies (IM- EDEA) that directed my attention back to my old unpub- lished results. 6 REFERENCES Allen H.K., Donato J., Wang H.H., Cloud-Hansen K.A., Davies J., Handelsman J. 2010. Call of the wild: antibiotic resis- tance genes in natural environments. Nature Rev. Micro- biol., 8: 251–259 Davis K.E.R., Joseph S.J., Janssen P.H. 2005. Effects of growth medium, inoculum size, and incubation time on culturabil- ity and isolation of soil bacteria. Appl. Environ. Microbiol., 71: 826–834 D’Costa V.M., McGrann K.M., Hughes D.W., Wright G.D. 2006. Sampling the Antibiotic Resistome. Science, 311: 374–377 Demaneche S., Sanguin H., Pote J., Navarro E., Bernillon D., Mavingui P., Wildi W., Vogel T.M., Simonet P. 2008. Antibi- otic resistance soil bacteria in transgenic plant fields. Proc. Nat. Acad. Sci., 105: 3957–3962 Hashimoto T., Hattori T. 1989. Grouping of soil bacteria by analysis of colony formation on agar plates. Biol. Fertil. Soils, 7: 198–201 Janssen P.H., Yates P.S., Grinton B.E., Taylor P.M., Sait M. 2002. Improved Culturability of Soil Bacteria and Isolation in Pure Culture of Novel Members of the Divisions Acidobac- teria, Actinobacteria, Proteobacteria, and Verrucomicrobia. Appl. Envir. Microbiol., 68: 2391–2396 Leps J., Smilauer P. 2003. Multivariate Analysis of Ecological Data using CANOCO. Cambridge, Cambridge University Press, UK: 280 p. Ley E.R., Lipson D.A., Schmidt S.K. 2001. Microbial biomass level in barren and vegetated high altitude talus soils. Soil Sci. Soc. Am. J., 65: 111–117 Nwosu V. 2001. Antibiotic resistance with particular reference to soil microorganisms. Res. Microbiol., 152: 421–430 Olsen R.A., Bakken L.R. 1987. Viability of soil bacteria: opti- mization of plate-counting technique and comparison be- tween total counts and plate counts within different size groups. Microb. Ecol., 13: 59–74 Riesenfeld C.S., Goodman R.M., Handelsman J. 2004. Uncul- tured soil bacteria are reservoir of new antibiotic resistance genes. Env. Microbiol., 6: 981–989 Stres B, Philippot L, Faganeli J, Tiedje J.M. 2010. Frequent freeze-thawcyclesyield diminishedyet resistant and respon- sive microbial communities in twotemperate soils: a labo- ratory experiment. FEMS Microb Ecol, doi:10.1111/j.1574- 6941.2010.00951.x Acta argiculturae Slovenica, 96/2, 87–93, Ljubljana 2010 doi:10.2478/v10014-010-0020-2 COBISS: 1.01 Agris category code: P10, T01 METHYLMERCURY INHIBITS GROWTH AND INDUCES MEMBRANE CHANGES IN Pseudomonas putida Maša VODOVNIK 1, Mirjana BISTAN 2, Maša ZOREC 3, Romana MARINŠEK LOGAR 4 Received October 08, 2010; accepted November 22, 2010. Delo je prispelo 08. oktobra 2010, sprejeto 22. novembra 2010. 1 Univ. of Ljubljana, Biotechnical Fac., Dept. of Animal Science, Groblje 3, SI-1230 Domžale, Slovenia, e-mail: masa.vodovnik@bf.uni-lj.si 2 Laboratory for environmental science and engineering, National institute of chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia 3 Same address as 1, Ph.D. 4 Same address as 1, Ph.D., e-mail: romana.marinsek@bf.uni-lj.si most toxic metals is certainly mercury, which represents a significant concern especially in aquatic ecosystems at sites with industrial or mining activities. Mining opera- tions in areas rich in cinnabar ore may represent strong sources of Hg for many years even after mining has been discontinued (Benoit et al., 1994). One of mercury (Hg) affected sites also lies in Western part of Slovenia where Methylmercury inhibits growth and induces membrane chang- es in Pseudomonas putida A bacterial model system (Pseudomonas putida DSM 50026) was used in this research to assess toxicity of the envi- ronmentally relevant concentrations of mercury species (MeHg and Hg(II)) that represent important pollutants of aquatic eco- systems at sites of industrial or mining activities. In addition to direct monitoring of bacterial growth, we also analyzed fatty acid profiles of exposed and non-exposed cultures to determine possible toxic effects manifested on membrane level. The results showed that exposure of P. putida to Hg(II) in concentrations of 0.2–200 µg/L did not have any significant effects on growth nor fatty acid composition of exposed bacterial culture. On the other hand, when bacteria were exposed to up to 1600-times lower concentrations of MeHg (0.12–12.5 µg/L), growth inhi- bition as well as significant changes in fatty acid composition were detected. Observed adaptive membrane changes due to MeHg exposure were similar to those associated with responses to organic solvents and some other membrane-disrupting com- pounds. Key words: microbiology / environmental protection / bacteria / Pseudomonas putida / aquatic ecosystems / pollution / mercury / methylmercury / growth inhibition / membrane adaptation / cis-trans isomerization Metil živo srebro inhibira rast in povzroča spremembe v mem- branah bakterije Pseudomonas putida V raziskavi smo na bakterijskem modelu (Pseudomonas putida DSM 50026) analizirali strupenost okoljskih koncentra- cij anorganske (Hg(II)) in organske (MeHg) oblike živega sre- bra, ki predstavljata pomembna vira onesnaženja vodnih eko- sistemov v bližini nekaterih industrijskih in rudarskih območij. Poleg neposrednega spremljanja bakterijske rasti smo analizi- rali tudi maščobnokislinske profile izpostavljenih bakterijskih kultur in jih primerjali s tistimi, ki živosrebrovima spojinama niso bili izpostavljeni. Rezultati so pokazali, da izpostavitev P. putida Hg(II) v koncentracijah med 0,2 in 200 µg/L ne inhibira rasti, niti ne vpliva na maščobnokislinsko sestavo bakterijskih membran. Nasprotno pa je izpostavitev celic do 1600-krat niž- jim koncentracijam MeHg povzročila tako upočasnitev rasti kot tudi prilagoditvene spremembe na membranskem nivoju. Slednje so bile podobne kot tiste, opažene ob izpostavitvi bakte- rij organskim topilom in nekaterim drugim spojinam, ki motijo integriteto membran. Ključne besede: mikrobiologija / varstvo okolja / bakte- rije / Pseudomonas putida / vodni ekosistemi / onesnaževanje / živo srebro / metil živo srebro / inhibicija rasti / membranska adaptacija / cis-trans izomerizacija 1 INTRODUCTION In the past few decades, environmental pollution has become one of the world’s major concerns. Heavy metals are a group of pollutants representing environ- mental problem in most parts of the world. One of the Acta agriculturae Slovenica, 96/2 – 201088 M. VODOVNIK et al. lies the second largest Hg mine in the world. The Idrija Mine operated for 500 years until its closure in 1994, but mercury laden tailings still line the banks and the sys- tem is a threat to the Idrija River and water bodies down- stream, including the Soča / Isonzo River and the Gulf of Trieste in the northern Adriatic Sea, which are therefore subjects to continuous environmental monitoring (Hines et al., 2000; Faganeli et al., 2003). Mercury can cause acute as well as chronic poison- ing in animals and humans (Sweet & Zelnikoff, 2001; Crespo-Lopez et al., 2007; Han et al., 2008). The most toxic forms of mercury are usually considered to be or- ganic compounds, such as methylmercury (MeHg) and dimethylmercury (Me2Hg), which have the tendency to accumulate in hydrophobic environments such as cell membranes (Mason et al, 1996; Girault et al., 1997). Bio- accumulation on different trophic levels leads to biomag- nification effect in natural ecosystems, which means that even low levels of organic mercury compounds in the en- vironment may have detrimental effect on organisms at the end of the food chain (Mason et al., 1996). However, not all the mercury that is present in natural ecosystems is bioavailable and therefore harmful to the living organ- isms (Golding et al., 2007). Standard chemical analytical methods do not have the power to discriminate between bioavailable and fixed forms of mercury in environmen- tal samples and therefore need to be complemented with methods, based on responses of living (micro)organisms for proper risk assessment (Farre et al. 2005). Living cells (organisms) can be used as bioindicators, as well as test- species in bioassays. By recent establishment of modern 3R concept (reduction, replacement, refinement), the development and application of bioassays based on mi- crobial cells is being promoted, due to their simple culti- vation in axenic cultures and lack of ethical issues usually present when using higher organisms (Marinšek Logar and Vodovnik, 2007). Cell membrane as the first barrier separating cel- lular interior from its environment represents a primary defense line against unfavorable environmental impacts and therefore appears to be a good target for ecophysi- ological as well as toxicological studies. Bacteria are known to react to several environmental triggers by modifying fatty acid composition of their membranes, predominantly by changing the ratio of saturated to un- saturated fatty acids (Cronan 2002). However, several strains of ubiquitous bacterium Pseudomonas putida have been shown to use at least three adaptation mecha- nisms at membrane level which apply to different types of environmental stressors: (1) changes in the overall de- gree of saturation of fatty acids (Loffhagen et al. 2004), (2) the formation of cyclopropane fatty acids (Härtig et al. 2005) and (3) cis-trans isomerization (von Wallbrun et al. 2003). Meanwhile the first two responses are mainly associated with temperature stress and starvation, cis- trans isomerization, appears to be involved in toxic stress defence (Heipieper et al. 1995; Heipieper et al. 1996). It has been shown in solvent-tolerant strains of P. putida that toxicity and concentration of organic solvents in their membrane correlate with increase in trans/cis fatty acid ratio (Heipieper et al. 1992; Heipiepper et al. 1994; Weber et al., 1996). Moreover, several heavy metal ions, namely Zn2+, Cd2+, Cr3+, Co2+, Cu2+ and Ni2+ have also been shown to induce adaptive changes resulting in in- creased accumulation of trans fatty acids (Heipieper et al. 1996). However, there are so far no reports on effects of mercury (Hg) species on P. putida (or other bacterial) membranes, which is the objective of this article. Our hy- pothesis was that organic (MeHg), as well as inorganic (Hg(II)) form of mercury may influence the membrane (fatty acid profile) of P. putida. However, due to its hy- drophobic nature, the degree to which membranes are affected was expected to be larger in case of MeHg. 2 MATERIALS AND METHODS 2.1 MICROORGANISM P. putida DSM 50026 cells were purchased in freeze- dried form from Deutsche Sammlung von Mikroorgan- izmen und Zellkulturen GmbH (DSMZ, 1998). 2.2 CHEMICALS General reagents dimethylsulfoxide (DMSO), methanol, n-hexane and glycerol as well as both mercury species: MeHg in the form of CH3HgCl and Hg(II) in the form of HgCl2 were purchased from Sigma (St. Luis, MO, USA) or Merck (Darmstadt, Germany). Bacteriological agar and peptone were purchased from Biolife (Milan, Italy) and meat extract from Becton Dickinson (New Jersey, NJ, USA). Standard calibration mixture of bac- terial FAME in methyl-caproate (BAME standard) was purchased from Sigma (St. Luis, MO, USA) and standard calibration mixture of bacterial FAMEs in hexane (MIDI standard) was purchased from Hewlett Packard (USA). 2.3 CULTURE CONDITIONS P. putida DSMZ 50026 was cultivated for 20 hours in medium described by DSMZ (1998) containing 3 g of meat extract and 5 g of bacteriological peptone per Acta agriculturae Slovenica, 96/2 – 2010 89 METHYLMERCURY INHIBITS GROWTH AND INDUCES MEMBRANE CHANGES IN Pseudomonas putida 1000 ml of distilled water (dH2O). Cells were grown in 10 ml test-tubes, at 27 °C (without shaking). 2.4 EXPOSURE CONDITIONS After 20 hours incubation, selected environmen- tally relevant concentrations (Quiu et al., 2006) of tested mercury species (Table 1) were added to the cultures. MeHg, which is water insoluble, was dissolved in 50% DMSO instead of distilled water before added to the cul- ture medium. Negative controls for those samples were performed with the addition of an adequate amount of DMSO as well. Cells exposed to tested concentrations of mercury species were incubated for another 24 hours at 27 °C. During incubations, growth was followed by measuring optical density at 654 nm by Novaspec II Visi- ble Spectrofotometer. Cells were harvested by centrifuga- tion (3000 rpm, 4 °C, 10 min). Pellets were resuspended in sterile double distilled water (1 mL), frozen (−20 °C) and freeze-dried. 2.5 LIPID EXTRACTION AND TRANSESTERIFI- CATION Bacterial lipids were extracted from freeze dried samples and transesterified using modified HCl/metha- nol procedure that has already been described before (Iv- ancic et al., 2009). 2.6 GAS CHROMATOGRAPHY Fatty acid methyl esters (FAMEs) extracts in hexane were analyzed on gas chromatograph Shimadzu GC-14A equipped with flame ionization detector (FID). Capillary column (Equity-1; Supelco, 28046-U) with non-polar stationary phase (100% poly-dimethyl-siloxane) was used. The analysis followed the temperature program: temperature gradient from 150 to 250 °C at 4 °C min−1. The flow rate of carrier gas (He) was 30 ml min−1. The injector temperature was held at 250 °C and detector at 280 °C. The results were registered on Chromatopac C-R6A integrator. Relative proportions of fatty acids between C10 and C20 were calculated from peak ar- eas. Identification was done either directly by compari- son of retention times of unknown peaks with standard fatty acid calibration mixtures (BAME, MIDI; SIGMA- Aldrich) or indirectly by equivalent chain length (ECL) factors calculation (Mjøs, 2003). Tested compound Concentrations (μg/L) HgCl2 (Hg (II)) 200 20 2 0.2 CH3HgCl (MeHg) 12.5 1.25 0.12 - Table 1: Concentrations of tested mercury species Preglednica 1: Koncentracije testiranih živosrebrovih spojin Figure 1: Growth curves of P. putida DSM 50026 culture exposed to MeHg in concentrations from 0.12 to 12.5 µg/L in comparison to non-exposed cells (control). The time of MeHg addition is marked by arrow. Slika 1: Rastne krivulje kulture P. putida DSM 50026 izpostavljene MeHg v koncentracijah od 0,12 do 12,5 µg/L v primerjavi s kon- trolno (neizpostavljeno) kulturo. Začetni čas izpostavitve MeHg je označen s puščico. Acta agriculturae Slovenica, 96/2 – 201090 M. VODOVNIK et al. 2.7 CALCULATIONS Membrane fatty acids which were present in less than 0.5% of total fatty acids were signed as fatty acids in traces and were not considered for further interpretation. Trans/cis ratio of unsaturated fatty acids was calcu- lated according to Heipieper et al., 1995. 2.8 STATISTICAL ANALYSIS All the exposures were performed in 4 parallel sam- ples. The data was statistically analyzed using Student`s t-test with significance level of 0.05. 3 RESULTS Our experiments showed that exposure of P. putida DSM 50026 to HgCl2 in concentrations of 0.2–200 µg/L did not result in any significant effects on growth nor fatty acid composition of exposed bacterial culture (re- sults not shown). However, when bacteria were exposed to organic mercury in form of CH3HgCl (from 0.12–12.5 µg/L) growth inhibition as well as significant changes in fatty acid composition were observed (Figures 1, 2). Most significant (dose-effect) inhibition of cell growth occurred in the first 4–6 hours after exposure of cells to MeHg. After 8 hours of growth in MeHg supplemented medium, cell culture appeared to grow with approxi- mately the same (attenuated) rate, regardless of mercury concentration. The behavior of growth curves that can be observed in Fig. 1 suggests the possibility of adaptive changes of microbial cells, enabling the culture to con- tinue growing under the changed conditions. Since the membrane represents the primary barrier between cells and the environment and is responsible to regulate the flow of molecules into (and out of) the cell, we decided to focus our research on possible adaptive changes that may be detected on lipid level. Our results show that exposure of P. putida to MeHg in concentrations between 0.12– 12.5 µg/L significantly influences fatty acid profile of tested bacteria, resulting in increase of trans/cis fatty acid ratio from 0.35 ± 0.04 (in non-exposed cells) to 0.47 ± 0.02 (in cells exposed to 0.12 µg/L or 1.25 µg/L MeHg) or 0.48 ± 0.03 (in cells exposed to 12.5 µg/L). The observed shift in trans/cis ratio is mainly associated with statisti- cally significant decreases in C16:1cis9 and C18:1cis11 fatty acids, accompanied by increase in C16:1trans9. 0,0 5,0 10,0 15,0 20,0 25,0 30,0 35,0 C 12 :0 C 12 :0 2 O H C 14 :0 C 15 :0 C 16 :1 c9 C 16 :1 t9 C 16 :0 C 17 :0 cy c C 18 :1 c1 1 C 18 :1 t1 1 C 18 :0 Fa tt y ac id % control 0,12 µg/L MeHg 1,25 µg/L MeHg 12, 5 µg/L MeHg Figure 2: Changes in fatty acid profile of P. putida DSM 50026 culture exposed to MeHg in concentrations from 0.12 to 12.5 µg/L. Slika 2: Spremembe v maščobnokislinskem profilu kulture P. putida DSM 50026 ob izpostavitvi MeHg v koncentracijah od 0,12 µg/L do 12,5 µg/L. Acta agriculturae Slovenica, 96/2 – 2010 91 METHYLMERCURY INHIBITS GROWTH AND INDUCES MEMBRANE CHANGES IN Pseudomonas putida 4 DISCUSSION Little is known about the molecular mechanisms controlling (methyl)mercury uptake and toxicity so far. The primary targets of both, CH3Hg(II) as well as in- organic Hg(II), are considered sulfhydril-containing macromolecules (especially of various molecular weight thiol-containing proteins). Covalent binding of mercury compounds to proteins acting as antioxidants (i.e. glu- tathione) or components of electron transport chains appears to be associated with free radical accumulation, leading to oxidative damage of macromolecules and lipid peroxidation (Patrick, 2002; Han et al., 2008). Despite the generally recognized common molecular targets, the levels of mercury species inducing toxicity usually differ. It is generally assumed that MeHg is the most toxic Hg species, which is often ascribed to its higher lipid solubil- ity (Sweet, 2001). However, the octanol/water partition coefficients (Kow) of uncharged HgCl2 and CH3HgCl spe- cies do not differ significantly (3.3 and 1.7 respectively) (Broniatowski, 2009). These data suggest that the actual interaction of the Hg species with the cell membranes is very much dependent on the environmental factors in- fluencing their ionization as well as membrane charge (especially pH and the types, as well as concentrations of ions present in the solution) (Sweet, 2001). Only few studies on methylmercury binding to bi- omembrane lipids have been reported. Early “in vitro’’ studies suggested a direct mechanism of CH3Hg(II) action on selected lipids. Segal & Wood (1974), for ex- ample, performed an NMR study which showed that MeHg can react both catalytically and directly with plasmalogens (a group of phospholipids which are im- portant in a membrane structure for cells of the cen- tral nervous system of higher organisms). They showed that MeHg ion is soluble in phospholipids and catalyses rapid hydration and hydrolysis of the vinyl ether linkage to give a mixture of palmitic and stearic aldehydes plus the linolenic monoglyceride product (Segal and Wood, 1974). Furthermore, studies performed by LeBlanc et al. (1984) revealed a pH-dependent binding of MeHg to acidic phosphatidylserine (PS) and phosphatidylinositol (PI) phospholipids, but not to zwitterionic phosphati- dilcholine (PC) and sphingomielin (SM). The most ex- tensive study on MeHg interaction with phospholipid membranes is probably the one performed by Girault et al.(1997), in which the authors used three complemen- tary approaches: (i) 199Hg-NMR which quantitatively describes MeHg mobility and complexation, both in so- lution and at the membrane interface, (ii) fluorescence polarization which reveals dynamic changes of the hy- drophobic interior and (iii) solid state 31P-NMR which is indicative of the phosphate group structure and mobility and allows detection of non-bilayer phases. The study revealed that CH3Hg(II) interactions with membrane li- pids are electrostatic in nature and primarily depend on the polar head groups negative charges (phosphate moi- ety), which is not the case with HgCl2(Delnomdedieum et al., 1992; Girault et al. 1997). Extensive metal binding (up to three MeHg molecules per lipid) induces limited membrane destabilization, which may, in some cases, be associated with loss of its integrity (Girault et al., 1997). Our results confirmed the hypothesis that effects of mercury compounds on P. putida cells essentially de- pend on their chemical structures. Meanwhile chosen concentrations of inorganic mercury in form of HgCl2, did not inhibit growth nor induced any adaptive changes in bacterial membranes the opposite was the case with its methylated form. MeHg exhibited toxicity that reflect- ed at both levels (culture growth as well as membrane changes) at concentrations up to 1600-times lower that the highest Hg(II) concentration tested. Most signifi- cant inhibition of cell growth occurred in the first 4–6 hours after exposure to MeHg. After 8 hours of growth in MeHg supplemented medium, cell culture appeared to grow with approximately the same (slightly attenuated) rate, regardless of methylmercury concentration. The lack of dose-effect inhibition at this stage may indicate that differences in chosen concentrations were too small to inhibit significantly different proportions of cells that would be observable by spectrophotometric measure- ments. However, the behavior of growth curves that can be observed in Fig. 1 suggests the possibility of adaptive changes in certain number of microbial cells that have survived the MeHg presence, enabling the cultures to continue growing under the changed conditions. Observed membrane changes associated with MeHg exposure resulted in overall increase in trans/cis fatty acid ratio, indicating the prevalent isomerization of cis- to trans- unsaturated fatty acids. This adaptive re- sponse is known to be associated with decrease in mem- brane fluidity, enabling Pseudomonas strains to grow in the presence of membrane-disrupting compounds (Von Wallbrun et al., 2003; Härtig et al., 2005). The same type of response has already been described when selected P. putida strains were exposed to toxic concentrations of toluene (Weber et al., 1994; Heipieper et al., 1994), phenol (Heipieper et al., 1992), ethanol (Heipieper et al., 1994) and six different heavy metals, namely Zn2+, Cd2+, Cr3+, Co2+, Cu2+ and Ni2+ (Heipieper et al., 1996). Th e de- gree of isomerization was shown to depend on the toxic- ity and the concentration of membrane-affecting agents. The described way of membrane adaptation is performed by cis–trans-isomerase (Cti), a constitutively expressed periplasmic enzyme that, to exert its action, necessitates neither ATP nor other cofactors, and consistently, is in- Acta agriculturae Slovenica, 96/2 – 201092 M. VODOVNIK et al. dependent of de novo synthesis of lipids. Due to its direct correlation with toxicity, cis–trans-isomerization is a po- tential biomarker for recording solvent stress or changes of other environmental conditions (Bernal et al., 2007; Heipieper et al., 2010). The question that needs to be ad- dressed at this point is how do P. putida cells detect the presence of membrane disrupting compounds like MeHg, which leads to activation of protective mechanism(s). In the presence of organice solvents, the detection and ac- tivation appears to be directly associated with detected increase in membrane microviscosity caused by changes of the acyl chain order (Killian et al., 1992). According to Neumann et al., the hydrophilic structure and peri- plasmic location of Cti supports the assumption that the enzyme can only reach its target (the double bonds of un- saturated fatty acids that are located at a certain depth of the membrane) when the membrane is destabilized (i.e. the fluidity at certain regions is increased) by environ- mental factors (Neumann et al., 2003; Härtig et al., 2005). Since direct effect on membrane fluidity has also been observed in the case of MeHg (Girault et al., 1997; Schara et al., 2001), the abovementioned mechanism may apply here as well. 5 CONCLUSIONS In our research a bacterial model has been used to as- sess toxicity of two mercury species that represent impor- tant pollutants of aquatic ecosystems at sites of industrial or mining activities. The results showed different toxicities of Hg(II) and MeHg to (bacterial) cells. Meanwhile inorganic form, Hg(II) did not influence the growth nor induce any signifi- cant changes in fatty acid profile of P. putida, exposure to methylated form of mercury resulted in partial growth in- hibition, which appears to be balanced by adaptive mem- brane changes. We showed that changes in fatty acid pro- file of P. putida resulting from MeHg exposure are similar to those observed as a response to organic solvents, as well as some other membrane-disrupting compounds, and are associated with (adaptive) decrease in membrane fluidity. Despite the fact that response of P. putida to MeHg is not specific, these bacteria might possibly be used to develop a bioassay, used to indicate the potential presence of toxic bioavailable concentrations of MeHg in environ- ments where mercury represents the major pollutant (i.e. Idrijca river, where MeHg also accumulates in freshwater fish and crabs). Neverthless, more research needs to be done to assess the influence of different physico-chemical parameters (like pH, ionic strength etc.) as well as other potentially interfering compounds on the responsiveness of the system. 6 REFERENCES Benoit G., Schwantes J.M., Jacinto G., Goudcollins M.R. 1994. Preliminary study of the redistribution and transformation of HgS from cinnabar mine tailings deposited in Honda Bay, Palawan, Philippines. Mar. Pollut. Bull., 28: 754–759 Bernal P., Segura A., Ramos J.L. 2007. Compensatory role of the cis-trans-isomerase and cardiolipin synthase in the mem- brane fluidity of Pseudomonas putida DOT-T1E. Environ. Microbiol., 9: 1658–1664 Broniatowski M., Dynarowicz-Łatka P. 2009. Search for the Molecular Mechanism of Mercury Toxicity. Study of the Mercury(II)−Surfactant Complex Formation in Langmuir Monolayers. J Phys Chem B., 113: 4275–4283 Crespo-López E.M., Lima de Sáa A., Herculanob A.M., Bur- banoc R.R., Martins do Nascimento H.L. 2007. Methylmer- cury genotoxicity: A novel effect in human cell lines of the central nervous system. Env. Internat., 33: 141–146 Cronan J.E. 2002. Phospholipid modifications in bacteria. Curr. Opin. Microbiol., 5: 202–205 Delnomdedieu M., Boudou A., Georgescauld D., Dufourc E.J. 1992. Specific interactions of mercury chloride with mem- branes and other ligands as revealed by mercury-NMR. Chem. Biol. Interact., 81: 243–69 DSMZ: German collection of microorganisms and cell cul- tures. Braunschweig, Deutsche Sammlung von Mikroor- ganismen und Zellkulturen GmbH. http://www.dsmz.de/bactnom/nam2400.htm (20. Oct. 2010) Faganeli J., Horvat M., Covelli S., Fajon V., Logar M., Lipej L., Cermelj B. 2003. Mercury and methylmercury in the Gulf of Trieste (northern Adriatic Sea). Sci. Total Environ., 304: 315–326 Farre M., Brix R., Barcelo D. 2005. Screening water for pollut- ants using biological techniques under European Union Funding during the last 10 years. Trends Anal. Chem. 24: 532–545 Girault L., Boudou A., Drfourc E.J. 1997. Methyl mercury in- teractions with phospholipid membranes as reported by fluorescence, 31P and 199Hg NMR. Biochim Biophys Acta, 1325: 250–262 Golding G.R., Kelly C.A., Sparling R., Loewen P.C., Barkay T. 2007. Evaluation of Mercury Toxicity as a Predictor of Mer- cury Bioavailability. Environ. Sci. Technol., 41: 5685–5692 Han Z.X., Lv C.X., Zheng Z.R. 2008. Toxicity of mercury to Carassius auratus and the antagonistic effect of selenium. Synth. React. Inorg. Met. Org. Chem., 38: 584–590 Härtig C., Loffhagen N., Harms H. 2005. Formation of trans Fatty Acids Is Not Involved in Growth-Linked Membrane Adaptation of Pseudomonas putida. Appl. Environ. Micro- biol., 71: 1915–1922 Heipieper H.J., Diefenbach R., Keweloh H. 1992. Conversion of cis unsaturated fatty acid to trans, a possible mechanism for the protection of phenol-degrading Pseudomonas putida P8 from substrate toxicity. Appl. Environ. Microbiol., 58: 1847–1852 Heipieper H.J., De Bont J.A.M. 1994. Adaptation of Pseudomo- nas putida S12 to ethanol and toluene at the level of fatty acid composition of membranes. Appl. Environ. Micro- biol., 60: 4440–4444 Acta agriculturae Slovenica, 96/2 – 2010 93 METHYLMERCURY INHIBITS GROWTH AND INDUCES MEMBRANE CHANGES IN Pseudomonas putida Heipieper H.J., Loffeld B., Keweloh H. 1995. The cis/trans isom- erisation of unsaturated fatty acids in Pseudomonas putida S12: An indicator for environmental stress due to organic compounds. Chemosphere, 30: 1041–1051 Heipieper H.J., Maulenbeld G., Van Oirschot Q., De Bont J.A.M. 1996. Effect of environmental factors on the trans/ cis ratio of unsaturated fatty acids in Pseudomonas putida S12. Appl. Environ. Microbiol., 62: 2773–2777 Heipieper H.J., Fischer J. and Meinhardt F. 2010. Cis–Trans Isomerase of Unsaturated Fatty Acids: An Immediate Bac- terial Adaptive Mechanism to Cope with Emerging Mem- brane Perturbation Caused by Toxic Hydrocarbons. In: Handbook of Hydrocarbon and Lipid Microbiology. Tim- mis K.N. (ed.). Springer Berlin Heidelberg: 1606–1602 Hines M.E., Horvat M., Faganeli J., Bonzongo J.C.J., Barkay T., Majorf E.B., Scott K., Bailey E.A., Warwick J.J., Lyons W.B. 2000. Mercury Biogeochemistry in the Idrija River, Slove- nia, from above the Mine into the Gulf of Trieste. Env. Res., 83 (A): 129–139 Ivancic T., Vodovnik M., Marinsek-Logar R., Stopar D. 2009. Conditioning of the membrane fatty acid profile of Esch- erichia coli during periodic temperature cycling. Microbiol- ogy, 155: 3461–3463 Killian J.A., Fabrie C.H., Baart W., Morein S., de Kruijff B. 1992. Effects of temperature variation and phenethyl alcohol ad- dition on acyl chain order and lipid organization in Esch- erichia coli derived membrane systems. A 2H- and 31P- NMR study. Biochim. Biophys. Acta, 1105: 253–262 Leblanc R.M., Joly L.P., Paiement J., 1984. pH-dependent inter- action between methyl mercury chloride and some mem- brane phospholipids. Chem. Biol. Interact., 48: 237–241 Loffhagen N., Härtig C., Babel W. 2004. Pseudomonas putida NCTC 10936 balances membrane fluidity in response to physical and chemical stress by changing the saturation de- gree and the trans/cis ratio of fatty acids. Biosci. Biotechnol. Biochem., 68: 317–323 Mason P.R., Reinfelder J.R., Morel F.M.M. 1996. Uptake, Toxic- ity, and Trophic Transfer of Mercury in a Coastal Diatom. Environ. Sci. Technol., 30: 1835–1845 Mjøs S.A. 2003. Identification of fatty acids in gas chromatog- raphy by application of different temperature and pressure programs on a single capillary column. J. Chromatogr., 1015: 515–161 Marinšek Logar R., Vodovnik M. 2007. The applications of mi- crobes in environmental monitoring. In: Communicationg Current Research and Educational Topics and Trends in Microbiology (Microbiology series, no. 1, vol. 2). Méndez– Vilas A. (ed.). Badajoz, Formatex: 577–585 Neumann G., Kabelitz N., Heipieper H.J. 2003. The regulation of the cis-trans isomerase of unsaturated fatty acids in Pseu- domonas putida: correlation between cti activity and K+- uptake systems. Eur. J. Lipid Sci. Technol., 105: 585–589 Patrick L. 2002. Mercury toxicity and antioxidants: Part 1: role of glutathione and alpha-lipoic acid in the treatment of mercury toxicity. Altern, Med. Rev., 7: 456–471 Qiu G., Feng X., Wang S., Shang L. 2006. Environmental con- tamination of mercury from Hg-mining areas in Wuchuan, northeastern Guizhou. China. Env. poll., 142: 549–558 Schara M., Nemec M., Falnoga I., Kobal A.B., Kveder M., Svetek J. 2001. The action of mercury on cell membranes. Cell. Mol. Biol. Lett., 6: 299–304 Segall H.J., Wood J.M. 1974. Reaction of methyl mercury with plasmalogens suggests a mechanism for neurotoxicity of metal-alkyls. Nature, 248: 456–458 Sweet L.E., Zelnikoff J.T. 2001. Toxicology and immunotoxicol- ogy of mercury: a comparative review in fish and humans. J. Toxicol. Environ. Health B Crit. Rev., 4 (B): 161–205 Von Wallbrunn A., Richnow H.H., Neumann G., Meinhardt F., Heipieper H.J. 2003. Mechanism of cis-trans isomerization of unsaturated fatty acids in Pseudomonas putida. J. Bacte- riol., 185: 1730–1733 Weber F.J., Isken S., De Bont A.M. 1994. Cis/trans isomeriza- tion of fatty acids as a defence mechanism of Pseudomonas putida strains to toxic concentrations of toluene. Microbiol- ogy, 140: 2013–2017 Acta argiculturae Slovenica, 96/2, 95–101, Ljubljana 2010 COBISS: 1.01 Agris category code: Q04 VARIATIONS IN THE FATTY ACID COMPOSITION AND NUTRI- TIONAL VALUE OF ADRIATIC SARDINE (Sardina pilchardus Walb.) THROUGH THE FISHING SEASON 1 Monika MARIN 2, Tomaž POLAK 3, Lea GAŠPERLIN 3, Božidar ŽLENDER 3 Received October 16, 2009; accepted August 30, 2010. Delo je prispelo 16. oktobra 2009, sprejeto 30. avgusta 2010. 1 This article is part of a M.Sc. thesis ‘ Lipid composition and sensory quality of the adriatic sardine (Sardina pilchardus) in different seasons ‘, issued by Monika Marin, supervisor Prof. Božidar Žlender, Ph.D. / Prispevek je del magistrskega dela Monike Marin z naslovom ‘Vpliv sezone ulova na lipidno sestavo in senzorično kakovost jadranske sardele (Sardina pilchardus)’, mentor prof. dr. Božidar Žlender 2 Droga Kolinska, Food Industry, d.d., Kolinska 1, SI-1000 Ljubljana, Slovenia 3 Univ. of Ljubljana, Biotechnical Fac., Dept. of Food Science and Technology, Jamnikarjeva 101, SI-1000 Ljubljana, Slovenia, e-mail: bozidar.zlender@bf.uni-lj.si Variations in the fatty acid composition and nutritional value of Adriatic sardine (Sardina pilchardus Walb.) through the fishing season We investigated the chemical composition, in terms of water, protein, ash, total fat and fatty-acid composition, of sardine meat, and estimated its nutritional value. The samples originated from Adriatic sardines (Sardina pilchardus Walb.) that were collected in the north Adriatic Sea through the win- ter, spring, summer and autumn seasons. The content of 20 fatty acids was determined by in-situ transesterification and capillary column gas-liquid chromatography, using nonadecanoic acid (19:0) as internal standard. The mean contents of the sardine meat were: 70.8% water, 21.0% protein, 2.5% ash and 6.4% fat. For the fatty-acid composition, 18.0% were mono-unsaturated, 42.6% polyunsaturated and 39.8% saturated. The total-fat con- tent increased through the year, from winter to autumn (0.69 to 18.15 g/100 g meat). The fatty-acid composition in the sardine meat varied significantly, with the levels of the polyunsaturated fatty acids (4.6 g/100 g meat), and especially eicosapentaenoic acid (20:5n-3, 0.98 g/100 g meat) and docosahexaenoic acid (22:6n-3, 1.9 g/100 g meat), being the highest in autumn, before spawning. The n-6/n-3 ratio (0.13) and P/S ratio (7.6) show that sardine meat can and should be included in a balanced human diet. Considering the recommended daily intake of n-3 polyun- saturated fatty acids is 0.45 g per day for a healthy population, this would be consumed as 10 g sardine meat collected in the autumn or 100 g sardine meat collected in the winter. Key words: human nutrition / food / fish / Adriatic sar- dine / Sardina pilchardus Walb. / composition / fatty acids / nu- tritional value / season Maščobnokislinski profil in prehranska vrednost jadranske sar- dele (Sardina pilchardus Walb.) v odvisnosti od sezone ulova Raziskovali smo kemijsko (vsebnost vode, beljakovin, maščob in mineralnih snovi) in še posebej maščobno kislin- sko sestavo mesa sardel ter določili njeno prehransko vrednost. Jadranske sardele (Sardina pilchardus Walb.) so bile ujete v se- vernem Jadranskem morju v štirih različnih lovnih sezonah: pozimi, spomladi, poleti in jeseni. Z metodo in situ transesteri- fikacije in določitve na plinsko tekočinskem kromatografu smo določili 20 maščobnih kislin. Meso sardin povprečno vsebu- je 70,8 % vode, 21,0 % beljakovin, 2,5 % mineralnih snovi in 6,4% maščob; od skupnih maščobnih kislin je 18,0 % enkrat nenasičenih, 42,6 % večkrat nenasičenih ter 39,8 % nasičenih. Vsebnost maščob je močno nihala med sezonami (naraščala od zime proti jeseni, od 0,69 % do 18,15 %) in statistično značil- no vplivala na maščobnokislinsko sestavo mesa. Največ večkrat nenasičenih maščobnih kislin (skupaj 4,6 g/100 g mesa), pred- vsem eikozapentaenojske (20:5n-3, 0,98 g/100 g mesa) in doko- zaheksaenojske (22:6n-3, 1,9 g/100 g mesa) so vsebovale sardele jesenskega ulova (pred drstenjem). Razmerja n-6/n-3 (0,13) in P/S (7,6) kažeta, da imajo sardine visoko prehransko vrednost. Priporočen dnevni vnos večkrat nenasičenih maščobnih kislin (0,45 g/dan za zdravo populacijo) dosežemo že z dnevnim za- užitjem 100 g mesa sardel, ujetih pozimi, in 10 g mesa sardel, ujetih jeseni. Ključne besede: prehrana ljudi / živila / ribe / jadranska sardela / Sardina pilchardus Walb. / sestava / maščobne kisline / prehranska vrednost / letni čas Acta agriculturae Slovenica, 96/2 – 201096 M. MARIN et al. 1 INTRODUCTION The Adriatic sardine (Sardina pilchardus Walb.) is a widespread fish along the European and African coasts of the Mediterranean Sea. It is also found in the English Channel, and in the Black Sea and North Sea. This pe- lagic oily fish was, and still is, an important species for the fish industry in Mediterranean countries. Evidence suggests that fish consumption decreases the risk of cardiovascular disease, cancers and asthma, and particularly consumption of the oily fish that contain high levels of polyunsaturated fatty acids (PUFAs), such as docosahexaenoic acid (DHA, 22:6cis-4,7,10,13,16,19) and eicosapentaenoic acid (EPA, 20:5cis-5,8,11,14,17). Fish consumption also has positive influences on infant neurodevelopment, along with many other reported ben- efits (Simopoulos, 2002; FSA, 2004). However, although all fish are generally considered to be of similar nutri- tional value, it has been recognized that the PUFA com- position can vary across different fish species. Therefore, when fish is suggested as a means of improving health, both the species and the PUFA composition should be taken into account (Osman et al., 2001). The role of lipids in nutrition is manifold. They are the carriers of the energy value and contain the essential FAs and PUFAs, and they are the precursors for the bio- synthesis of the eicosanoids, which have important func- tions in the human body (Tapiero et al., 2002; Nelson and Cox, 2005). Indeed, the most important value of fish meat and lipids are the PUFAs, and especially DHA and EPA, both of which are found mainly in fish lipids and in trace amounts in food of animal and plant origin. Studies of the FA composition of fish are important when related to the above-mentioned benefits. From the nutritional point of view, there are different important indices, such as the P/S ratio, n-6/n-3 ratio and athero- genic index. The recommended ratio of PUFA to satu- rated FAs (P/S) should be above 0.4, with the normal P/S ratio for meat at around 0.1 (Wood et al., 2003). Simopoulos (2002) concluded that the optimal ratio of n-6/n-3 varies from 1:1 to 4:1, depending on disease un- der consideration. A low ratio of n-6/n-3 PUFAs is more desirable to reduce the risks of many of chronic diseases of high prevalence in both Western societies and devel- oping countries, which are being exported to the rest of the World. The World Health Organisation has recom- mended similar values for the n-6/n-3 PUFA ratio, as 5:1 to 10:1 (WHO, 1994). Attempts to develop a better index of the potential health attributes of foods contain- ing a mixture of FAs have been reported by Ulbricht and Southgate (1991), with their indices of atherogenicity and thrombogenicity. The atherogenic index is recom- mended as between 0.70 and 0.72 (Ulbricht and South- gate, 1991), although Salobir (1997) recommended even lower levels, of under 0.5. In view of these details, we have here followed the need for further studies of the nutritional value and lipid profiles of the most commonly consumed fish in Medi- terranean countries throughout the four seasons of the year. 2 MATERIAL AND METHODS The sardines (Sardina pilchardus) used in this study were caught in the north Adriatic Sea, off Slovenia and north Croatia, through the four seasons of the year: win- ter, spring, summer and autumn. Three samples were taken during each season, each of which consisted of eight sardines. The fish were prepared as ready-to-eat: they were beheaded, gutted and frozen (−80 °C), for a maximum period of one month before sampling. The chemical composition of these sardines was de- termined at the end of the sample collection, in terms of their ash, water, proteins and fat content. The water content was determined on 5 g samples of the minced meat. These samples were dried in an oven at 105 °C ac- cording to the AOAC 950.46 (1997) recommendations. The total protein (crude protein, N × 6.25) content was determined by the Kjeldahl method, according to the AOAC 928.08 (1997) standard methodology. The ash content was determined by mineralisation of the samples at 550 °C, according to the AOAC 920.153 (1997) stand- ard method. The FA compositions of the samples were deter- mined by gas-liquid chromatography (GLC). The meth- od chosen was in-situ transesterification (Park and Goins, 1994). The contents of the FA methyl esters (FAMEs) were determined by GLC using an Agilent Technologies 6890 gas chromatograph with a flame ionisation detector and an SPTM-2380 capillary column (Supelco, Cat.No. 24111) (60 m × 0.25 mm × 0.2 mm). The separation and detec- tion were performed under the following conditions: temperature programme, 170 °C (hold 5 min); 4 °C/min to 250 °C (25 min); injector temperature, 250 °C; detec- tor temperature, 280 °C; injector: split:splitless, 1:30, vol- ume 1 mL; carrier gas: He, 1.0 mL/min; make-up gas: N2, 45 mL/min; detector gases: H2, 40 mL/min; synthetic air (21% O2) 450 mL/min. The FAMEs were determined through comparison with the retention times of the FAMEs from standard mixtures (Supelco, 37-component FAME ester mix; Cat. No. 18919-1AMP; Supelco, PUFA No.1: Animal source, Cat. No. 47015-U; Supelco, Linoleic Acid Methyl Ester cis/trans Isomer Mix, Cat. No. 47791; Supelco, cis-7-octa- decenoic methyl ester, Cat. No. 46900-U; Supelco, cis-11- Acta agriculturae Slovenica, 96/2 – 2010 97 VARIATIONS IN THE FATTY ACID COMPOSITION AND NUTRITIONAL VALUE ... THROUGH THE FISHING SEASON octadecenoic methyl ester, Cat. No. 46904; Fluka, methyl stearidonate, Cat. No. 43959; natural ASA CLA 10t, 12c in CLA 9c, 11t; NuChek standards: GLC-68D, GLC-85, GLC-411 and GLC-546). As an internal standard, 100 mL of a solution of nonadecanoic acid in hexane (10 g/L; C19:0) (Sigma, N4129) was added to the samples be- fore saponification. The NuChek GLC-68D and GLC-85 standard mixtures were used to determine the response factor, Rfi, for each FA. The weight portion of each FA in the sample was determined using Rfi and the transforma- tion factor of FA content from the FAME content. The reliability and accuracy of the analytical methods for the detection of the FAs was ensured by the use of the certi- fied CRM 163 reference matrix (blended beef-pork fat; BCR), and these were in good agreement with the certi- fied values. The FAs are expressed in % of total FAs (w/w) and as mg FA in 100 g of edible fish. The data for the chemical composition of sardine meat were analysed by least squares analysis using the GLM procedure (SAS, 1999). The statistical model in- cluded the effects of season (S) and fish (F): yijk = m + Si + Fj + eijk, where yijk = the ijk th observation, m = the general mean, Si = the effect of the i th catching season (winter, spring, summer, autumn), Fj = the effect of the j th fish (1–24 fish), and eijk = the residual random term with a variance of σ 2 e. The means of the experimental groups were ob- tained using the Duncan test, with relationships between the parameters assessed by Pearson correlation coeffi- cients, using the CORR procedure (SAS, 1999). 3 RESULTS AND DISCUSSION The fat content and the FA profiles of the sardine meat varied significantly according to the seasons of cap- ture (P ≤ 0.001). Table 1 gives the mean values for the chemical compositions, which are similar to other data reported in the literature (Castrillón et al., 1997; Mac- ciola et al., 2003). The ash content was higher (2.5%) in comparison with other meats. The main reason for this was in the preparation of the samples, as they were pre- pared ready-to-eat: beheaded and gutted, but without the fish bones being removed as they are edible after cook- ing. The amount of total fat varied over the period of a year, in the range of 0.7% to 18.2% (Table 2). Changes in fat content vary as the sardines drain or replenish their fat reserves in response to the availability of food, their spawning cycles and other factors in the sea (Hardy and Keay, 1972). Adriatic sardines spawn from the end of autumn to the end of winter, when little food is available, and therefore their fat stores are used up during this period. The sardines accumulate fat in their tissues for spawning and wintering in temperate seasons (sum- mer, autumn), as is seen by the data given in Table 2. Zlatanos and Laskaridis (2007) reported that sar- dines collected during winter had the highest lipid con- tent (10.6%). Our findings are not in agreement with their results, whereby our data show that the sardines collected at the end of winter, and at the end of spring to the begin- ning of summer, had the lowest fat content (Table 2). The highest fat content was seen in sardines collected in late summer and autumn, and the values were much higher than those reported for the Greek study (8.3% to 15.1%, vs. 5.9% to 8.5%). Similar results for the fat content of marinated sardines (highest in late summer and autumn, Chemical composition, % N Mean ± SD Min Max CV (%) Protein Beljakovine 96 20.96 ± 0.98 19.62 22.63 4.68 Water Voda 96 70.79 ± 5.54 61.05 76.62 7.82 Fat Maščobe 96 6.42 ± 6.16 0.69 18.15 95.98 Ash Pepel 96 2.50 ± 0.25 2.13 2.83 9.86 Table 1: Chemical composition of the sardine meat Preglednica 1: Kemijska sestava mesa sardin N – number of observations / število vzorcev, – mean / povprečje, SD – standard deviation / standardni odklon, Min. – minimal value / minimalna vrednost, Max. – maximal value / maksimalna vrednost, SD – standard deviation / standardni odklon, CV (%) – coefficient of variation / koeficient variabilnosti. Parameter, % Winter /Zima Spring / Pomlad Summer / Poletje Autumn / Jesen P-value / p-vrednost Fat / Maščoba 0.97 ± 0.22c 1.26 ± 0.43c 8.33 ± 2.38b 15.11 ± 3.08a < 0.0001 Table 2: Fat content (%) of the sardine meat (N = 24) with respect to catching season Preglednica 2: Vsebnost maščob (%) v mesu sardin v odvisnosti od sezone ulova Mean values ± standard deviation. N – number of sardines in each season. a,b,c − mean values of seasons with different letters are statistically signifi- cant different (P < 0.05) Povprečne vrednosti ± standardni odklon. N – število vzorcev v vsaki sezoni. a,b,c − srednje vrednosti z različnimi nadpisanimi črkami se statistično značilno (p ≤ 0.05) razlikujejo. Acta agriculturae Slovenica, 96/2 – 201098 M. MARIN et al. and lowest in winter and spring) were reported by Mac- ciola et al., (2003), while Hardy and Key (1972) saw lower levels of fat during spawning, due to the fat mobilisation associated with gametogenesis. The FA composition of the fish investigated in the present study are given in Table 3. The data show remark- able and significant changes (P ≤ 0.05 or less) in the indi- vidual FAs during this one-year period. On average, the sardine meat contained 6.4% fat, and for the FA composition, 18.0% was mono-unsatu- FA / % total FAs MK / % skupnih MK Winter Zima Spring Pomlad Summer Poletje Autumn Jesen Mean ± SD Povprečje ± so P-value p-vrednost 12:0 0.03 ± 0.01c 0.06 ± 0.01b 0.07 ± 0.01b 0.08 ± 0.00a 0.07 ± 0.01 < 0.0001 14:0 1.42 ± 0.07c 5.67 ± 1.03b 6.68 ± 0.30ab 7.26 ± 0.37a 5.94 ± 1.65 < 0.0001 15:0 0.46 ± 0.01c 0.87 ± 0.09b 0.83 ± 0.06b 1.01 ± 0.04a 0.86 ± 0.15 < 0.0001 16:0 23.13 ± 0.76c 27.48 ± 1.37a 24.73 ± 1.18bc 26.30 ± 1.07ab 26.24 ± 1.85 0.0057 16:1cis-9 1.36 ± 0.04c 4.45 ± 1.07ab 4.18 ± 0.14b 5.39 ± 0.21a 4.37 ± 1.22 0.0009 17:0 0.79 ± 0.01c 0.95 ± 0.06b 1.09 ± 0.04a 1.14 ± 0.07a 1.01 ± 0.12 < 0.0001 18:0 6.08 ± 0.39a 4.99 ± 0.36b 4.92 ± 0.24b 5.79 ± 0.30a 5.24 ± 0.53 0.0002 18:1trans-9 0.13 ± 0.03c 0.14 ± 0.02c 0.25 ± 0.01b 0.36 ± 0.03a 0.22 ± 0.09 < 0.0001 18:1cis-9 4.97 ± 0.01d 6.87 ± 0.62c 11.77 ± 1.54b 14.38 ± 0.90a 9.59 ± 3.56 < 0.0001 18:2cis-9,12 2.62 ± 0.02b 3.03 ± 0.34a 2.49 ± 0.07b 2.99 ± 0.10a 2.86 ± 0.33 0.0028 18:2 CLA 0.96 ± 0.01c 1.99 ± 0.16b 2.17 ± 0.04a 2.04 ± 0.05ab 1.96 ± 0.32 < 0.0001 18:3cis-9,12,15 0.73 ± 0.00c 2.39 ± 0.28b 3.52 ± 0.13a 3.74 ± 0.06a 2.83 ± 0.89 < 0.0001 20:1cis-11 0.50 ± 0.13d 2.73 ± 0.23c 4.03 ± 0.25a 3.12 ± 0.19b 2.95 ± 0.91 < 0.0001 22:1cis-13 1.82 ± 0.17a 0.91 ± 0.11b 0.59 ± 0.11c 0.96 ± 0.15b 0.92 ± 0.32 < 0.0001 18:4cis-6,9,12,15 0.85 ± 0.17c 1.16 ± 0.08a 1.04 ± 0.04ab 0.95 ± 0.03bc 1.06 ± 0.13 0.0027 20:5cis-5,8,11,14,17 (EPA) 6.94 ± 0.07b 8.92 ± 1.23a 7.78 ± 0.66ab 7.23 ± 0.63b 8.12 ± 1.21 0.0692 22:3cis-13,16,19 0.62 ± 0.04c 1.63 ± 0.15a 1.20 ± 0.11b 1.03 ± 0.09b 1.31 ± 0.35 < 0.0001 22:4cis-10,13,16,19 1.52 ± 0.03a 0.95 ± 0.13b 0.77 ± 0.10bc 0.74 ± 0.08c 0.90 ± 0.23 < 0.0001 22:5cis-7,10,13,16,19 1.04 ± 0.06a 1.03 ± 0.07a 1.03 ± 0.11a 1.13 ± 0.12a 1.05 ± 0.10 0.2041 22:6cis-4,7,10,13,16,19 (DHA) 44.03 ± 0.56a 23.77 ± 4.40b 20.88 ± 2.61b 14.39 ± 1.39c 22.50 ± 8.04 < 0.0001 SFA / nasičene MK 31.92 ± 1.24c 40.02 ± 2.31ab 38.32 ± 1.59b 41.58 ± 1.79a 39.36 ± 3.11 0.0007 MUFA / enkrat nenasičene MK 8.78 ± 0.27d 15.10 ± 1.50c 20.82 ± 1.76b 24.20 ± 1.20a 18.03 ± 4.88 < 0.0001 PUFA / večkrat nenasičene MK 59.30 ± 0.96a 44.88 ± 3.50b 40.87 ± 3.23b 34.22 ± 1.99c 42.60 ± 7.12 < 0.0001 ΣEPA+DHA 50.97 ± 0.63 32.69 ± 5.63 28.66 ± 3.27 21.62 ± 2.02 30.62 ± 9.25 < 0.0001 Σn-3 55.73 ± 0.93a 39.85 ± 3.82b 36.31 ± 3.32b 29.20 ± 2.07c 37.78 ± 7.46 < 0.0001 Σn-6 3.58 ± 0.03c 5.02 ± 0.32a 4.66 ± 0.10b 5.02 ± 0.11a 4.83 ± 0.46 < 0.0001 n-6/n-3 0.07 ± 0.00c 0.13 ± 0.02b 0.13 ± 0.01b 0.17 ± 0.02a 0.13 ± 0.03 < 0.0001 P/S 9.23 ± 0.28a 7.26 ± 0.29c 7.28 ± 0.21c 7.92 ± 0.29b 7.57 ± 0.62 < 0.0001 IA 0.43 ± 0.02b 0.86 ± 0.11a 0.86 ± 0.06a 0.98 ± 0.07a 0.85 ± 0.16 < 0.0001 Table 3: Fatty acid composition (% of total fatty acid) of the sardine meat (N=24) with respect to catching season Preglednica 3: Maščobnokislinski profil (% od skupnih maščobnih kislin) maščobe v mesu sardin v odvisnosti od sezone ulova Mean value ± standard deviation. N – number of sardines in each season. a, b, c, d − mean values of seasons with different letters are statistically significant different (p < 0.05). Σn-3 − sum of 18:3cis-9,12,15, 18:4cis-6,9,12,15, 20:5cis-5,8,11,14,17, 22:3cis-13,16,19, 22:4cis-10,13,16,19, 22:5cis- 7,10,13,16,19 and 22:6cis-4,7,10,13,16,19. Σn-6 – sum of 18:2cis-9,12 and 18:2 CLA. IA − index of atherogenicity = (12:0 + 4 × 14:0 + 16:0)/(Σ(n-6) + Σ(n-3) + 18:1cis-9 + other MUFA) (Ulbricht and Southgate, 1991). Povprečje ± standardni odklon. N – število vzorcev v vsaki sezoni. a,b,c,d − srednje vrednosti z različnimi nadpisanimi črkami se statistično značilno (p ≤ 0.05) razlikujejo. Σn-3 − vsota 18:3cis-9,12,15, 18:4cis-6,9,12,15, 20:5cis-5,8,11,14,17, 22:3cis-13,16,19, 22:4cis-10,13,16,19, 22:5cis- 7,10,13,16,19 in 22:6cis-4,7,10,13,16,19. Σn-6 – vsota 18:2cis-9,12 in 18:2 CLA. IA − indeks aterogenosti = (12:0 + 4 × 14:0 + 16:0)/(Σ(n-6) + Σ(n-3) + 18:1cis-9 + druge MUFA) (Ulbricht in Southgate, 1991). Acta agriculturae Slovenica, 96/2 – 2010 99 VARIATIONS IN THE FATTY ACID COMPOSITION AND NUTRITIONAL VALUE ... THROUGH THE FISHING SEASON rated, 42.6% polyunsaturated, and 39.4% saturated. The lipids of this sardine meat contained large proportions of palmitic acid (16:0; 26.2%) and DHA (22.5%). Oleic acid (18:1cis-9; 9.6%), EPA (8.1%), myristic acid (14:0; 5.9%), stearic acid (18:0; 5.2%), palmitooleic acid (16:1cis-9; 4.4%), α-linolenic acid (18:3cis-9,12,15; 2.8%) and do- cosapentaenoic acid (22:5cis-7,10,13,16,19; 1.1%) were all present as minor components. The saturated FAs in the sardine fat ranged from 31% to 45%; the highest proportions being seen as pal- mitic (26.2%) and myristic (5.9%) acids. In spring, sum- mer and autumn, the fat had a significantly higher weight percentage in the 16:0 palmitic acid than in winter (P < 0.05), presumably due to the spawning season and wintering. The content of the 14:0 myristic acid increas- ing during the year by more than five-fold, with its peak in autumn. In our study, these weight percentages of the 14:0 and 16:0 FAs were more variable in comparison with the values reported by Bandarra et al. (1997) and Zlatanos and Laskaridis (2007), which were seen to be particularly constant throughout the year and did not appear to be influenced by the diet of the sardines. This phenomenon could be explained according to the sea temperature and its oscillations: in the north Adriatic Sea, the tempera- tures oscillate from 8 °C in winter to 29 °C in summer and at the beginning of autumn. More food is available in the warmer periods, which the sardines accumulate as fat for the colder periods. In warmer seas, food is more uniformly available and therefore less fat accumulation is needed, which would explain the more constant satu- rated FA composition seen by others. The mono-unsaturated FAs in the lipids ranged from 8.8% to 24.2%; with the highest proportions seen for oleic acid (18:1n-9C). Here, these 18:1n-9C levels (5.0% to 14.4%) were similar than those reported by Bandarra et al. (1997) in their Portugese study (7.4% to 14.3%) and Zlatanos and Laskaridis (2007) in a Greek study (3.5% to 10.6%). The PUFAs in the sardine lipids ranged from 34.2% to 59.3%, with the highest proportions seen for DHA (mean, 22.5%) and for EPA (mean, 8.1%). The DHA con- tent decreased from winter (44%) to autumn (14%), and generally the EPA content increased significantly during spring and summer, when compared to the winter-au- tumn period (8.9% and 7.8% vs. 6.9% and 7.2%, respec- tively). Our data for the n-3 PUFAs are in agreement with the literature (Bandarra et al., 1997), although they were higher (mean, 37.8%) than those reported by Zlatanos and Laskaridis (2007) in the Greek study (35.3%) and much higher than those reported by Luzia, Sampaio, Cas- tellucci, and Torres (2003) in the Brazilian study (13.4% in summer). In the present study, there was a negative correlation between the fat content and that of the n-3 PUFAs: the n-3 PUFAs were low during the months with a high fat content (R = −0.84, P = 0.0001) (not presented in tables). In contrast, the saturated FA content increased during the months with a high fat content, in summer and autumn (R = 0.81, P = 0.0001). These data are in agreement with other studies (Bandarra et al., 1997; Macciola et al., 2003; Zlatanos and Laskaridis, 2007). From the nutritional point of view, the P/S ratio, n-6/n-3 ratio and the atherogenic index were calculated. The P/S ratio was high (7.6 in average) due to presence of large content of DHA (22:6cis-4,7,10,13,16,19) and EPA (20:5cis-5,8,11,14,17) (Table 3). The n-6/n-3 ratio in the present study (mean, 0.13) was favourable mainly be- cause of the low n-6 FA content. This ratio is considered to be a risk factor in cancers and coronary heart disease, and it is recommended that it is less than 10.0, and even less than 4.0 (WHO, 1994; Simopoulos, 2002; FSA, 2004). The introduction of a correct combination of sar- dines and other food into the diet can assure a good balance for human nutrition. On the other hand, in the present study, the atherogenic index values of the sardine varied, from 0.43 to 0.98 (mean, 0.85), which is higher than recommended (lower than 0.72). The main reason for these high atherogenic index values appears to lie in the presence of 14:0 myristic acid. The myristic acid value according to the atherogenic index is enhanced be- cause of the large influence of cholesterol in the blood (Ulbricht and Southgate, 1991). Aside from the relatively high atherogenic index, there can be further benefits from the high amounts of n-3 PUFAs that are seen, be- cause they balance the n-6/n-3 ratio, which in a modern western diet is generally greater than 15:1 to 16.7:1 (Si- mopoulos, 2002). The Food Standards Agency (FSA, 2004) has pub- lished recommendations for the daily intake of n-3 PU- FAs: 0.45 g daily for the protection of the adult popula- tion. Chapkin (1992) recommends 0.8 g of EPA and DHA daily for a healthy adult population, while Simopoulos (2002) recommends 0.65 g of EPA and DHA daily (cal- culated on a 8,400 kJ diet). This should be increased two- fold or more for the n-3 PUFA intake for a population with cardiovascular disease: from 0.9 g to 1.5 g of n-3 PUFAs (FSA, 2004), and up to 2 g to 4 g of EPA and DHA daily (AHA, 2003). As can be seen from the present study, a healthy adult can satisfy their daily n-3 PUFA intake (0.45 g; FSA, 2004) by eating only 10 g of the sardine meat collected in the autumn, or 100 g of the sardine meat when the sardines are lean. Similarly, people with different cardio- vascular diseases can cover the recommended amounts of the n-3 PUFAs by eating just 30 g of the sardine meat collected in autumn, and up to 330 g of the sardine meat Acta agriculturae Slovenica, 96/2 – 2010100 M. MARIN et al. collected in spring. Also, for the DHA and EPA intake for a healthy population, 20 g of the sardine meat collected in autumn or 220 g of the sardine meat collected in win- ter would cover the daily recommendations, with higher amounts of the sardine meat covering the EPA and DHA recommendations for cardiovascular patients. The full data for the analysis of the individual FAs (expressed as mg FA/100 g edible sardine meat) are given in Table 4. 4 CONCLUSIONS Although n-3 PUFAs are not only found in fish, as they are also present in flax seeds and nuts in particular, the EPA and DHA from fish fat are much more efficiently incorporated into the human body. α-linolenic acid, as a short n-3 PUFA, must be converted in the body to DHA and EPA, a process that is not particularly efficient in many people (and especially not so in the elderly), thus indicating that the direct consumption of DHA and EPA FA / mg FA/100 g meat MK / mg MK/100 g mesa Winter Zima Spring Pomlad Summer Poletje Autumn Jesen Mean Povprečje 12:0 0 ± 0 1 ± 0 5 ± 0 11 ± 0 4 ± 1 14:0 12 ± 1 64 ± 12 501 ± 23 988 ± 50 343 ± 95 15:0 4 ± 0 10 ± 1 63 ± 5 137 ± 5 50 ± 9 16:0 202 ± 7 312 ± 16 1855 ± 88 3578 ± 145 1516 ± 107 16:1cis-9 12 ± 0 51 ± 12 314 ± 10 733 ± 28 252 ± 71 17:0 7 ± 0 11 ± 1 82 ± 3 155 ± 9 58 ± 7 18:0 53 ± 3 57 ± 4 369 ± 18 787 ± 41 303 ± 30 18:1trans-9 1 ± 0 2 ± 0 18 ± 1 49 ± 5 12 ± 5 18:1cis-9 43 ± 0 78 ± 7 883 ± 115 1956 ± 122 554 ± 206 18:2cis-9,12 23 ± 0 34 ± 4 187 ± 5 406 ± 13 165 ± 19 18:2 CLA 8 ± 0 23 ± 2 163 ± 3 277 ± 7 113 ± 19 18:3cis-9,12,15 6 ± 0 27 ± 3 264 ± 10 509 ± 8 164 ± 52 20:1cis-11 4 ± 1 31 ± 3 302 ± 19 424 ± 25 170 ± 53 22:1cis-13 16 ± 1 10 ± 1 44 ± 8 130 ± 20 53 ± 19 18:4cis-6,9,12,15 7 ± 1 13 ± 1 78 ± 3 129 ± 4 61 ± 7 20:5cis-5,8,11,14,17 (EPA) 61 ± 1 101 ± 14 584 ± 49 983 ± 85 469 ± 70 22:3cis-13,16,19 5 ± 0 19 ± 2 90 ± 8 140 ± 13 76 ± 20 22:4cis-10,13,16,19 13 ± 0 11 ± 2 58 ± 8 100 ± 11 52 ±13 22:5cis-7,10,13,16,19 9 ± 1 12 ± 1 77 ± 8 154 ± 16 61 ± 6 22:6cis-4,7,10,13,16,19 (DHA) 385 ± 5 270 ± 50 1566 ± 196 1957 ± 189 1300 ± 464 SFA / nasičene MK 279 ± 11 455 ± 26 2874 ± 119 5656 ± 244 2275 ± 179 MUFA / enkrat nenasičene MK 77 ± 2 172 ± 17 1561 ± 132 3291 ± 163 1042 ± 282 PUFA / večkrat nenasičene MK 519 ± 8 510 ± 40 3065 ± 243 4654 ± 270 2462 ±412 ΣEPA+DHA 446 ± 6 371 ± 64 2150 ± 245 2940 ± 274 1769 ± 534 Σn-3 487 ± 8 453 ± 43 2723 ± 249 3971 ± 282 2183 ± 431 Σn-6 31 ± 0 57 ± 4 349 ± 8 683 ± 16 279 ± 26 SFA/100 g meat / SMK/100 g mesa 874 ± 23 1137 ± 134 7500 ± 581 13602 ± 797 5778 ± 1273 g fat/100 g meat / g maščobe/100 g mesa 0.97 ± 0.03 1.26 ± 0.15 8.33 ± 0.65 15.11 ± 0.89 6.42 ± 1.41 Table 4: Mean FA levels (mg FA/100 g meat) in the sardine meat (N = 24) with respect to catching season Preglednica 4: Povprečne vrednosti maščobnih kislin (mg FA/100 g mesa) v mesu sardine (N = 24) v odvisnosti od sezone ulova Mean value ± standard deviation. N – number of sardines in each season. Σn-3 − sum of 18:3cis-9,12,15, 18:4cis-6,9,12,15, 20:5cis-5,8,11,14,17, 22:3cis-13,16,19, 22:4cis-10,13,16,19, 22:5cis-7,10,13,16,19 and 22:6cis-4,7,10,13,16,19. Σn-6 – sum of 18:2cis-9,12 and 18:2 CLA. Povprečje ± standardni odklon. N – število vzorcev v vsaki sezoni. Σn-3 − vsota 18:3cis-9,12,15, 18:4cis-6,9,12,15, 20:5cis-5,8,11,14,17, 22:3cis- 13,16,19, 22:4cis-10,13,16,19, 22:5cis-7,10,13,16,19 in 22:6cis-4,7,10,13,16,19. Σn-6 – vsota 18:2cis-9,12 in 18:2 CLA. Acta agriculturae Slovenica, 96/2 – 2010 101 VARIATIONS IN THE FATTY ACID COMPOSITION AND NUTRITIONAL VALUE ... THROUGH THE FISHING SEASON is preferable. In conclusion, due to the increasing impor- tance of the n-3 FAs for our health, the aim of this study was achieved: the definition of the FA composition of sardine collected in the Adriatic sea throughout the year, to provide this specific information for food specialists to include sardines in their menus for different kind of diets. The differences in the FA composition through the different periods of the year should also be taken into ac- count in these diets. 5 REFERENCES AHA. 2003. American Hearth Association recommendations: New guidelines focus on fish, fish oil, omega-3 fatty acids. Circulation: Journal of the American Hearth Association. http://www.americanhearth.org/presenter. jhtml?identifier=3006624 AOAC. 1997. Official Methods of Analysis. 16th ed. Washing- ton: Association of Official Analytical Chemists Bandarra N.M., Batista I., Nunes M.L., Empis J.M., Christie W.W. 1997. Seasonal changes in lipid composition of sar- dine (Sardina pilchardus). Journal of Food Science, 62: 40−42 Castrillón A.M., Navarro P., Alvárez-Pontes E. 1997. Changes in chemical composition and nutritional quality of fried sardine (Clupea pilchardus) produced by frozen storage and microwave reheating. Journal of the Science of Food and Agriculture, 75: 125−132 Chapkin R.S. 1992. Reappraisal of the essential fatty acids. In: Fatty Acids in Foods and their Health Implications. New York, Marcel Dekker, Inc.: 557−568 FSA. 2004. Advice on fish consumption: benefits and risks. Lon- don: Food Standards Agency and Department of Health. http://www.sacn.gov.uk Hardy R., Keay J.N. 1972. Seasonal variations in the chemical composition of Cornish mackerel, Scomber scombrus (L) with detailed reference to the lipids. Journal of Food Tech- nology, 7: 125−137 Luzia L.A., Sampaio, G.R., Castellucci, C.M.N., Torres, E.A.F.S. 2003. The influence of season on the lipid profiles of five commercially important species of Brazilian fish. Food Chemistry, 83: 93−97 Macciola V., de Leonardis A., Nichilo A. 2003. Studio degli acidi grassi omega-3 nelle sardine marinate (Sardina pilchardus). Industrie Alimentari, 42: 849−854 Nelson D.L., Cox, M.M. 2005. Principles of Biochemistry. 4th edition. New York, W. H. Freeman: 1119 p. Osman H., Suriah A.R., Law E.C. 2001. Fatty acid composition and cholesterol content of selected marine fish in Malaysian waters. Food Chemistry, 73: 55−60 Park P.W., Goins R.E. 1994. In-situ preparation of fatty acid methyl esters for analysis of fatty acid composition in foods. Journal of Food Science, 59: 1262–1266. Salobir K. 1997. Physiological role of meat in balanced nutri- tion. In: Meat in Nutrition and Health. Ljubljana, Biotech- nical Faculty: 161–170 SAS Software. 1999. Version 8.01. Cary, SAS Institute, Inc. Simopoulos A.P. 2002. The importance of the ratio of omega-6/ omega-3 essential fatty acids. Biomedicine & Pharmaco- therapy, 56: 365−379 Tapiero H., Nguyen Ba G., Couvreur P., Tew K.D. 2002. Poly- unsaturated fatty acids (PUFAs) and eicosanoids in human health and pathologies. Biomedicine & Pharmacotherapy, 56: 215−222 Ulbricht T.L.V., Southgate D.A.T. 1991. Coronary heart disease – seven dietary factors. Lancet, 338: 985–992 WHO. 1994. Fats and oils in human nutrition. In: Report of Joint Expert Consultation. Rome, WHO, Food and Agri- culture Organization of the United Nations: 147 p. Wood J.D., Richardson R.I., Nute G.R., Fisher A.V., Campo, M.M., Kasapidou, E., Sheard, P.R., Enser, M. 2003. Effects of fatty acids on meat quality: a review. Meat Science, 66: 21–32 Zlatanos L., Laskaridis K. 2007. Seasonal variation in the FA composition of three Mediterranean fish: sardine (Sardina pilchardus), anchovy (Engraulis encrasicholus) and picarel (Spicara smaris). Food Chemistry, 103: 725–728 Acta argiculturae Slovenica, 96/2, 103–109, Ljubljana 2010 COBISS: 1.01 Agris category code: L01 FORAGING BEHAVIOUR OF SHEEP AT PASTURE WITH DIFFERENT TYPES OF VEGETATION IN A PADDOCK 1 Manja ZUPAN 2, Danijela BOJKOVSKI 3, Ivan ŠTUHEC 4, Dragomir KOMPAN 5 Received October 06, 2010; accepted November 24, 2010. Delo je prispelo 06. oktobra 2010, sprejeto 24. novembra 2010. 1 Prispevek je del diplomske naloge Danijele Tomažič, mentor prof. dr. Ivan Štuhec, somentor prof. dr. Dragomir Kompan The article is a part of graduation thesis issued by Danijela Tomažič, supervisor Prof. Ivan Štuhec, Ph.D., co-advisor Prof. Dragomir Kompan, Ph.D. 2 Univ. of Ljubljana, Biotechnical Fac., Dept. of Animal Science, Groblje 3, SI-1230 Domžale, Slovenia, Assist., Ph.D., M.Sc., e-mail: manja.zupan@bf.uni-lj.si 3 The same address as 1, M.Sc., e-mail: danijela.bojkovski@bf.uni-lj.si 4 The same address as 1, Prof., Ph.D., M.Sc., e-mail: ivan.stuhec@bf.uni-lj.si 5 The same address as 1, Assoc. Prof., Ph.D., M.Sc., e-mail: drago.kompan@bf.uni-lj.si Foraging behaviour of sheep at pasture with different types of vegetation in a paddock This experiment was designed to study the foraging be- haviour of ewes on a pasture with paddocks with three dif- ferent types of vegetation, herbaceous (n = 3), woody (n = 2), and semi open (n = 1). Forty sheep were bred on a farm in the Karst region of Slovenia. Ten sheep were focally observed dur- ing day light (5 a.m.–9 p.m.). Ewes were observed for 2 days in each paddock with one rotation, so 12 days in total. Grazing time, circadian rhythm of grazing, drinking frequency, and fre- quency of salt consumption were the observed behaviours. On average, ewes grazed for 10.5 hours a day (mean ± SD = 626.2 ± 47.2 min), with a significant difference between individual variation  (P < 0.001). Sheep grazed the most in herbaceous paddocks (P < 0.001), with lower yet similar levels observed in woody and semi open paddock. The frequency of drinking and salt consumption was low. Individual grazing sheep would drink slightly less than once per day, while consuming salt on average 1.25 times per day. Drinking frequency was the highest in the semi open paddock with some trees and bushes, whereas salt consumption was most frequently observed in the woody paddocks. Key words: sheep / animal behaviour / ethology / graz- ing / pastures / paddocks / vegetation / Karst / drinking / salt consumption / Slovenia Obnašanje ovc na kraškem pašniku z različno vegetacijo Poskus je bil zastavljen z namenom proučiti obnašanje ovc na pašniku s čredinkami, v katerih so obstajale tri različne vrste vegetacije: travna ruša (n = 3), gozdna (n = 2) in delno za- raščena z drevesi in grmovjem (n = 1). Trop 40 ovc se je pasel na kmetiji na kraškem svetu v Sloveniji. Deset individualnih ovc je bilo direktno opazovanih v času dnevne osvetlitve (od 5. do 21. ure). Ovce so bile opazovane 2 zaporedna dneva v posamezni čredinki z eno ponovitvijo, torej skupaj 12 dni. Opazovana je bila dolžina zauživanja zelinja (+ listje iz grmovja in dreves), dnevni ritem zauživanja travne ruša, pogostost pitja in zauživa- nja soli. Na dan so se ovce pasle povprečno 10,5 ur (povprečje ± SD = 626,2 ± 47,2 min), vendar so bile značilne razlike med ovcami (P < 0,001). Najdaljši čas za pašo so ovce imele v čredin- kah s travno rušo (P < 0,001). Podoben čas paše je bil opazovan v gozdnih čredinkah in pol odprti čredinki. Pogostost pitja in zauživanja soli je bila nizka. Živali so pile malo manj kot enkrat na dan, medtem ko so zaužile sol 1,25 krat na dan. Pogostost pitja je bila največja v pol odprti čredinki, kjer so bila prisotna tudi drevesa in grmovje, medtem ko je bila največja pogostost zauživanja soli v gozdnih čredinkah. Ključne besede: ovce / obnašanje živali / etologija / paša / pašniki / čredinke / vegetacija / Kras / pitje / zauživanje soli / Slovenija Acta agriculturae Slovenica, 96/2 – 2010104 M. ZUPAN et al. 1 INTRODUCTION Small ruminant social environment varies widely from very intensive feeding with no grazing opportuni- ties to more extensive grazing areas with high grazing opportunities. Behavioural constraints are different and more diversified at pasture than indoors. As the human population becomes more aware of the quality of food, maximising forage utilisation through grazing is an in- creasingly important tool in animal production. Allow- ing the animals to perform more natural behaviour while being outside in the open can also improve their welfare (Špinka, 2006). In Slovenia, sheep production is the most wide- spread form of extensive animal production. In Karst, western part of Slovenia that occupies almost half of the territory (SURS, 2002), 85% of the countries sheep and goats are reared (SURS, 2002). This area is difficult to cul- tivate by agricultural machinery and is classified as inap- propriate for agricultural production. In 1998, between 120.000 and 150.000 ha of agricultural land in the Karst region was abandoned and overgrown with shrubs, trees, and brushwood (Cunder, 1998). Woody plants are a common component of the overgrowth. However, small ruminants can keep the pastures and farm land clear of ingrown woods through their capacity to graze. Small ruminants can also contribute to the safeguarding of agricultural functions, like care and preservation of the landscape, through maintaining grasslands and prevent- ing land from bush encroachment and fires. Grazing is defined as the time spent each day in graz- ing activity, that is, prehension and mastication (Wood- ward, 1997). It is well documented that sheep and goats show selective grazing and select for a high quality, nu- tritionally balanced diet. Grazing duration and rhythm is often related to specific forage characteristics (Baumont et al., 2000) due to different dietary choices (Morand-Fe- hr, 2003). This is partly the reason for keeping sheep and goats together at pasture. They have different preferences for feeds and the area is therefore more intensely grazed. At pasture, two main grazing periods usually occur at sunrise and sunset, which are also the preferred drinking times in both sheep (Rook and Penning, 1991) and goats (Rossi and Scharrer, 1992). However, drinking frequency can differ greatly between individuals, partly as a result of differences in social hierarchies (Milinski and Parker, 1991) and space availability around the drinking troughs (Ehrlenbruch et al., 2010). In a heterogeneous environment, the management of the grazing circuit has become an important factor. An understanding of sheep behaviour in a complex environ- ment is therefore essential for optimizing the manage- ment of sheep and goat flocks in unfavourable areas, such as the Karst region in Slovenia. Studying more feeding behaviour of ruminants may provide a firm knowledge on ethological traits of animals and a better understand- ing of how to achieve a good economical production, good animal welfare, and at the same time preserve the semi open landscape as best as possible. The aim of the study was to investigate how forage characteristics and type of vegetation influence foraging behaviour during sheep grazing. Sheep were observed at three different types of vegetation in a paddock in the Karst region of Slovenia. 2 MATERIAL AND METHODS 2.1 MATERIAL The experiment was carried out during the sum- mer time on a farm in the hilly Karst region of Slovenia called Vremščica (altitude of 900 – 1000 m a.s.l.). Two days before the onset of an experimental procedure, 40 ewes of Istrian Pramenka and 10 goats cross breeds were mixed and released into the same foraging area in order to get familiar with each other. The animals were reared on that farm and thus used to the area. The ani- mals were of similar age to prevent any effects of age on different flocking behaviour and handling responses (Hargreaves and Hutson, 1997). The area was fenced and for the purpose of the experiment divided into 6 pad- docks of similar size (approx. 400 m2). The shape of pad- docks depended on the structure of the ground and the type of vegetation. There were 3 paddocks covered with only grass sward (herbaceous paddock), 1 partly covered with trees and bushes (semi open paddock) and 2 fully overgrown paddocks with hazel and beech trees (woody paddock) (Fig. 1). Thus, 50 animals were grazing in six paddocks with three different types of vegetation for a period of 6 weeks. Animals stayed at the pasture for 24 hours. At each paddock there was one drinking trough and one wooden trough with salt. Therefore, animals had the possibility to both feed and drink. The average ambi- ent temperature during the observation period of 12 days was 14.1 °C, ranging from 9.2 °C to 18.5 °C. 2.2 METHODS 2.2.1 BEHAVIOURAL OBSERVATIONS For the purpose of this experiment ten young sheep were directly observed during foraging in the flock of 50 animals. Each of the 10 ewes was marked with a differ- ent combination of stripes on the back using red, black, Acta agriculturae Slovenica, 96/2 – 2010 105 FORAGING BEHAVIOUR OF SHEEP AT PASTURE WITH DIFFERENT TYPES OF VEGETATION IN A PADDOCK or green water resistant colour spray. The observations lasted 12 days during a 6 week period, with six inter- rupted intervals each lasting two consecutive days. Ani- mals were 5 or 6 days at a paddock, depending on the feeds availability, and afterwards moved to another pad- dock. During the experiment the animals were rotated between all 6 paddocks in such a way that they were ob- served twice at the same paddock. Two observers, situ- ated on a raised platform, began to observe the animals two days after moving the sheep into a paddock. Animals were observed inside (woody paddock) or outside the paddock (herbaceous paddock, semi open paddock). When observed outside, observers sat in a caravan 300 m distanced from the pasture, using binoculars. Before the observations started in the woody paddocks, the observers had spent 2 days at the pasture together with the sheep, so that the animals got used to their presence. The observers always wore the same working coat that was familiar to the animals. Observers started with the observations on the third day after moving the animals into a specific paddock. Observations started at 5 a.m. and finished at 9 p.m. Only one observer per time was observing the animals, 2 hours in a row, and then the observers were changed. Daily observation time was 16 hours. Activities of an individual sheep were scored on sheets of paper. Recordings were made for the following foraging activities: – grazing (duration, daily rhythm), – drinking water, – salt consumption. Grazing was recorded every 5 min during 16 hours of observation using instantaneous sampling. The number of drinking and salt consumption bouts was scored within the same time period but using continuous sampling. 2.2.2. STATISTICAL ANALYSIS We prepared data with Microsoft Excel for Win- dows and analysed them using statistical package SAS/ STAT (SAS, 2008). The general linear model (GLM) was used to determine the effects of normally distributed data. The daily values of data were tested for normality. All the tests were two-tailed and the significant level was set at P≤0.05. Data for grazing was normally distributed and the model shown in equation [1] was developed us- ing three fixed effects and an independent variable. The effect of breed had been tested but later omitted from the model as it described only 0.000126% of the variability. yijk = µ +Pi+ Dj + Ak + bi(tijkl – t –) + eijk, (1) where Pi is the effect of paddock (i = 1–6), Dj is the ef- fect of day (j = 1–2), Ak is the effect of animal (k = 1–10), bi(tijkl – t –) is the effect of averaged daily temperature, and eijk is a random error. In the case of other activities, drinking frequency and salt intake, data were not normally distributed even after transformation. The number of drinking and salt consumption bouts was very low; therefore, these behav- iours are presented in a descriptive manner only. 3 RESULTS 3.1 GRAZING Sheep were free at pasture for 24 hours and it was observed that during the afternoon heat sheep moved into the shade, if available. This suggests shade should be made available to animals in pastures. At pasture, sheep could develop their own foraging strategy. They spent on average 10.5 hours grazing during light hours (mean ± SD = 626.2 ± 47.2 min). The maximum duration of graz- a) b) c) Figure 1: Paddock with three different types of vegetation (a: herbaceous paddock, b: semi open paddock with some trees and bushes, c: woody paddock). Slika 1: Čredinke z različno vegetacijo (a: travnata čredinka, b: delno zaraščena čredinka z grmovjem in drevesi, c: gozdna čredinka). Acta agriculturae Slovenica, 96/2 – 2010106 M. ZUPAN et al. ing per day was a bit less than 12 hours (713.3 min). The similar duration of grazing was observed in the study of Lynch et al. (1992). They reported 8–9 hours of grazing a day with a maximum of around 13 hours when the feed supply was limited. This means that the broad diversity of feeds in our study motivated ewes to graze. The be- ginning of the grazing was synchronous in our study. If one animal started to graze, it was followed by the others. Such behaviour is species specific, and with sheep being social animals, they tend to be synchronous in their start- ing of grazing bouts (Champion et al., 1994). The circadian rhythm of grazing was significantly different during the day (Fig. 2). Grazing was the most intense in the morning with the peak between 6–8 a.m. when animals would spend from 45–60 min per hour grazing, whereas in the afternoons between 6–7 p.m., animals would graze during the entire observational pe- riod. After 5 p.m., the amount of time grazing increased with animals spending more than 50% of their time on that activity. This trend was also reported by Shinde et al. (1997) where grazing was generally observed at any time of day or night, but was most intensive in the morn- ing and late afternoon until dusk. Lynch et al. (1992) ex- plained that in continental areas grazing activity is con- centrated to 4 hours after dawn and in the last 4 hours around sunset, but can easily start before dawn and ex- tend long into the dark. As shown in Fig. 3, we found significant differ- ences in the grazing duration in different types of pad- dock (P < 0.001). Sheep spent the most time grazing at the herbaceous paddock whereas at the woody and semi open paddock the grazing was reduced to a similar level. This means that soil (grass) and aerial (woody) feeding behaviour differed. According to Vidrih et al. (1996) and Baumont et al. (2000), differences in herbage composi- tion between types of paddock can affect grazing dura- tion, and may explain the differences observed in our study. There was an additional effect of the individual on grazing duration (P < 0.001; Fig. 4). The variation in the average grazing duration over the 12 observed days ranged between 592 and 662 min. The temperature did not significantly affect the graz- ing time (P > 0.1), but it affected the circadian rhythm of grazing (P < 0.05; Fig. 5). When temperature was below 15.4 °C, sheep grazed more during 9 a.m. and 4 p.m. than when above 15.4 °C. This is a predictable result since ru- minants tend to avoid grazing during the hottest part of the day and thus reduce their daily grazing time. To avoid thermal stress ruminants find shade and spend more time resting (Shinde et al., 1997). Figure 2: Daily grazing rhythm at herbaceous paddock, woody, and semi open paddock with some trees and bushes. Slika 2: Dnevni ritem paše v travni čredinki, gozdni čredinki in delno zaraščeni čredinki z grmovjem in drevesi. 625 617 637 605 610 615 620 625 630 635 640 herbaceous woody semi open Type of paddock M in ut es p er d ay * Figure 3: Grazing time in different types of paddock. Difference between bars: F2,60 = 5.23; *P < 0.001. Slika 3: Čas paše v rezličnih vrstah čredink. Razlika med stolpci: F2,60 = 5,23; *P < 0.001. Acta agriculturae Slovenica, 96/2 – 2010 107 FORAGING BEHAVIOUR OF SHEEP AT PASTURE WITH DIFFERENT TYPES OF VEGETATION IN A PADDOCK On the first observed day, sheep spent less time graz- ing compared to the second day (F1,60  = 122.33; P < 0.001; Table 1). Sheep thus showed different foraging strategies between days due to shortages in herbage availability during the second day. This is based on the conclusions of Baumont et al. (2000), where it is suggested that one of the limiting factors of grazing is the herbage availability and growth stage of vegetation. 3.2 DRINKING Animals had free access to water. On the basis of visual observations, it can be stated that ewes approached the water trough very suddenly and they would always run towards it. It was observed that when one or two animals started to approach the watering point, the other animals followed. According to this, we support the con- clusion that drinking behaviour by sheep is socially fa- cilitated (Forkman, 1996) and a synchronised behaviour (Rook and Penning, 1991), with similar findings in Vid- rih et al. (1996). As all the animals arrived at the drinking source at approximately the same time, competition for water was most probably high (Ehrlenbruch et al., 2010). When sheep are housed indoors, an increased number of ewes per nipple drinker may lead to an increase in total drinking time and number of displacements (Bøe, 1998). On some days some ewes were not observed to drink during the observational period. However, Lynch et al. (1992) concluded that during summer sheep should drink at least once a day, otherwise they tend to reduce grazing time in the heat and increase grazing at night or early in the morning when dew is on the grass. Accord- ing to our results, it can be concluded that in the case of a small flock of grazing sheep, enough space for drink- ing water should be available at the pasture, so that the 560 580 600 620 640 660 680 1 2 3 4 5 6 7 8 9 10 Animal Th e du ra tio n of gr az in g * Th e du ra tio n of gr az in g Figure 4: The average grazing time of an individual sheep. Dif- ference between bars: F9,60 = 4.58; *P < 0.001. Slika 4: Povprečen čas paše za posamezno žival. Razlika med stolpci: F9,60 = 4,58; *P < 0.001. Figure 5: Grazing rhythm at temperatures above and below the average daily temperature. Difference between the lines: F1,60 = 4.08; P < 0.05. Slika 5: Pašni ritem nad in pod povprečno dnevno temperaturo. Razlika med nad in pod povprečno dnevno temperaturo: F1,60 = 4.08; P < 0.05. Behaviour Day F-value P-value1 2 Grazing time 589.25 ± 33.75 663.1 ± 23.99 122.33 < 0.001 Table 1: Differences in the grazing duration between 2 observed days (variables are given as mean min ± SD) Preglednica 1: Razlike med časom paše v dveh opazovanih dneh (spremenljivke so podane kot min ± SD) Acta agriculturae Slovenica, 96/2 – 2010108 M. ZUPAN et al. majority of animals have access to the water at the same time. The drinking frequency was low during the obser- vation period (Table 2). The average drinking frequency during this time was 0.99 per animal. The maximum number of drinking bouts per observation period was two. The animals drank the most frequently in the morn- ing between 8 a.m. and 9 a.m., and between 3 p.m. and 7 p.m. in the afternoon. The water usage differed be- tween the types of paddock. Sheep drank the most in the semi open paddock, but in the herbaceous and the woody paddock the frequency was lower, yet the same 3.3 SALT CONSUMPTION Salt appetite or sodium hunger is a motivational state in which animals seek out and ingest substances contain- ing sodium (Johnson and Thunhorst, 1997). Sheep in our study had access to feed on leaves from bushes and trees. The expected result is that the frequency of salt intake was the highest at the woody paddock, with a lower value for the herbaceous paddock. At the semi open paddock animals were not seen to consume salt during the obser- vation period (Fig. 7). Salt consumption occurred mainly in the morning between 6 a.m. and 8 a.m., and between 5 p.m. and 9 p.m. in the afternoon. The average frequency of salt consumption per day was 1.25 (Table 2). However, individual variation existed between the animals for salt consumption (Fig. 8). This may show that sheep differ in their taste preference of feed found in the environment (Baumont et al., 2000). Vidrih et al. (1996) analysed the concentration of particular minerals in the leaves of ha- zel and beech tree, with leaves containing 0.15–1.17 g so- 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 1 2 3 4 5 6 7 8 9 10 Animal Fr eq ue nc y of dr in ki ng Fr eq ue nc y of dr in ki ng Figure 6: Frequency of drinking of an individual sheep. Slika 6: Pogostnost pitja pri posamezni ovci. 0.8 0.8 1.5 0.5 3.1 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 herbaceous w oody semi open Type of paddock Fr eq ue nc y of a ct iv iti es Figure 7: Frequency of drinking (grey bar) and salt consumption (black bar) in paddock with different types of vegetation. Slika 7: Pogostnost pitja (sivi stolpec) in konzumacija soli (črn stolpec) v čredinkah z različno vegetacijo. Observation days Drinking Salt intake 1 1.1 0.0 2 1.0 0.1 3 0.9 0.0 4 0.9 0.0 5 1.4 2.2 6 1.4 2.5 7 0.8 1.3 8 0.4 0.5 9 2.1 4.5 10 0.9 3.1 11 0.3 0.3 12 0.7 0.5 Mean 0.99 1.25 Table 2: Drinking frequency and salt intake frequency between the observation days Preglednica 2: Pogostnost pitja in konzumacije soli med opazo- vanimi dnevi (Fig. 7). Climate conditions affected drinking behaviour as well. When the temperature was higher, there was a greater need for water. Water usage was different among animals, showing genetic influence on the behaviour ex- pressed. The animal that drank the most often was one of the two ewes that spent the most time grazing. The lowest frequency among ewes was 0.5 and observed by animal 8. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 1 2 3 4 5 6 7 8 9 10 Fr eq ue nc y of sa lt co ns um pt io n Animal Fr eq ue nc y of sa lt co ns um pt io n Figure 8: Frequency of salt consumption of an individual sheep. Slika 8: Pogostnost konzumacije soli pri posamezni ovci. Acta agriculturae Slovenica, 96/2 – 2010 109 FORAGING BEHAVIOUR OF SHEEP AT PASTURE WITH DIFFERENT TYPES OF VEGETATION IN A PADDOCK dium/kg of dry matter. Sheep weighing less than 50 kg have a nutritional requirement of 1.5 g of sodium in dry matter per day to maintain optimum health (Vidrih et al., 1996). It was observed that animals would chew the bark off trees or wood at the woody paddock. This might be a consequence of the lack of sodium (Kermauner, 1996). However, further study of the nutritional value of forages is required. 4 CONCLUSIONS For sheep, time spent grazing was on average 10.5 hours per day during the light hours of 5 a.m. to 9 p.m. The type of paddock influenced the grazing duration and daily rhythm. The frequency of water drinking was over- all low with animals drinking less than once per day. The highest water usage was recorded at the semi open pad- dock. It can be concluded that enough space for drink- ing should be available on the pasture, especially at semi open paddock, since sheep are showing synchronised drinking behaviour. The frequency of salt consumption was the highest at the woody paddock, which can be ex- plained by the lack of sodium in the leaves and branches that are often eaten. It is advised to provide additional so- dium, in the form of salt, under such environmental con- ditions. In conclusion, foraging behaviour under grazing conditions is greatly influenced by differences between individual ewes and forage conditions. 5 ACKNOWLEDGEMENTS The authors would like to thank Tom Fraser for help with the English and acknowledge Andreja Komprej and Špela Malovrh for their assistance with the statistical analysis. 6 REFERENCES Baumont R., Prache S., Meuret M., Morand-Fehr P., 2000. How forage characteristics influence behaviour and intake in small ruminants: a review. Livest. Prod. Sci., 64: 15–28 Bøe K.E. 1998. Drikkeadferd for drektige søyer, med fokus på antall dyr per drikkenippel (Drinking behaviour of preg- nant ewes, with focus on number of animals per water nipple). ITF report no. 93. Ås, Norges Landbrukshøgskole, Institutt for tekniske fag: 9 p. Champion R.A., Rutter S.M., Penning P.D., Rook A.J. 1994. Temporal variation in grazing behaviour of sheep and re- liability of sampling periods. Appl. Anim. Behav. Sci., 42: 99–108 Cunder T. 1998. Zaraščanje kmetijskih zemljišč in ukrepi za preprečevanje opuščanje pridelave (Overgrowing of ag- ricultural area and concepts for the prevention of cultiva- tion). Yearly Report 1998. Ljubljana, Ministry of science and technology, Republic of Slovenia: 59 p. Ehrlenbruch R., Pollen T., Andersen I.L., Bøe K.E. 2010. Com- petition for water at feeding time – The effect of increasing number of individuals per water dispenser. Applied Animal Behaviour Science, 126: 105–108 Forkman B., 1996. The social facilitation of drinking: what is facilitated, and who is affected? Ethology, 102: 252–258 Hargreaves A.L., Hutson G.D. 1997. Handling systems for sheep. Livestock production science, 49: 121–138 Johnson A.K., Thunhorst R.L. 1997. The neuroendocrinol- ogy of thirst and saltappetite: visceral sensory signals and mechanisms of central integration. Front. Neuroendocri- nol. 18: 292– 353 Kermauner A. 1996. Prehrana in krma za drobnico (Nutrition and feed for small ruminants) In: Reja drobnice (Small ruminant breeding). Kompan D. (ed.). Ljubljana, ČZD Kmečki glas: 77–135 Lynch J.J., Hinch G.N., Adams D.B. 1992. The behaviour of sheep. Biological principles and implications for produc- tion. Walingford, CSIRO Publications: 237 p. Milinski M., Parker G.A., 1991. Competition for resources. In: Behavioural Ecology. Krebs J.R., Davies N.B. (eds.). Oxford, Blackwell Scientific: 137–168 Morand-Fehr. P. 2003. Dietary choices of goats at the trough. Small Ruminant Research, 49: 231–239. Rook A.J., Penning P.D. 1991. Synchronization of eating, ru- minating and idling activity by grazing sheep. Appl. Anim. Behav. Sci. 32: 157–166 Rossi R., Scharrer E. 1992. Circadian patterns of drinking and eating in pygmy goats. Physiol. Behav., 51: 895–897 SAS. 2008. Statistical Analysis Systems. Version 9.1. Inc., Cary, NC, USA Shinde A.K., Karim S.A., Patnayak B.C. Mann J.S. 1997. Dietary preference and grazing behaviour of sheep on Cenchrus cili- aris pasture in semi arid region of India. Small Ruminant Research, 26: 119–122 SURS. 2002. Statistical office of the Republic Slovenia: 659 Špinka M. 2006. How important is natural behaviour in animal farming systems? Applied Animal Behaviour Science, 100: 117–128 Vidrih T., Kompan D., Kermauner A., Pogačnik M., Kotar M., Kotnik T. 1996. V kolikšni meri paša na kraški ruši lahko zadovolji prehranske potrebe drobnice. V: Zbornik. Možnosti razvoja reje drobnice v Sloveniji, Postojna, No- vember. Slovenj Gradec, Kmetijska založba: 39–44 Woodward S.J.R. 1997. Formulae for predicting animals’ daily intake of pasture and grazing time from bite weight and composition. Livestock production science, 52: 1–10 Acta argiculturae Slovenica, 96/2, 111–115, Ljubljana 2010 COBISS: 1.01 Agris category code: M40 FLUCTUATING ASYMMETRY IN DIPLOID FEMALE AND STERILE TRIPLOID RAINBOW TROUT (Oncorhynchus mykiss) Jurij POHAR 1, Klavdija STRGAR 2 Received November 05, 2010; accepted November 26, 2010. Delo je prispelo 05. novembra 2010, sprejeto 26. novembra 2010. 1 Univ. of Ljubljana, Biotechnical Fac., Dept. of Animal Science, Groblje 3, Domžale, SI-1230, Slovenia, Assist. Prof., Ph.D., M.Sc., e-mail: jure.pohar@bf.uni-lj.si 2 The same address as 1 Fluctuating asymmetry in diploid female and sterile triploid rainbow trout (Oncorhynchus mykiss) Viability of an organism and possibility to survive in natu- ral environment could be judged by the magnitude of fluctu- ating asymmetry (FA) which is defined as random deviation from perfect symmetry of an organism. In order to estimate if there is the difference in FA between diploid female and sterile triploid rainbow trout (Oncorhynchus mykiss) the number of rays in pelvic and pectoral fins was determined on both sides of body in 150 individuals from two populations which were of the same genetic origin and were reared under same farm conditions. Units of asymmetry were determined as the abso- lute value of difference between counts on both sides of body. Results indicate that diploids exhibit larger FA than triploids in both traits; however the difference between both populations is statistically significant only if the number of units of asym- metry for both traits for each fish is summed up. The need to estimate the viability of these two populations on the basis of other traits is discussed and the necessity to use the metric traits to determine FA is stressed out. Key words: fish / rainbow trout / Oncorhynchus mykiss / viability / fluctuating asymmetry / developmental stability / sterile triploids / diploid females Fluktuacijska asimetrija pri diploidnih in sterilnih triploidnih samicah kalifornijske postrvi (Oncorhynchus mykiss) Vitalnost nekega organizma in verjetnost, da preživi v naravi, je mogoče presojati na podlagi velikosti fluktuacijske asimetrije (FA). Ta je določena s tem, koliko telo določenega osebka odstopa od popolne simetrije. Da bi ocenili ali se sami- ce in sterilni triploidni osebki kalifornijske postrvi (Oncorhyn- chus mykiss) med seboj razlikujejo v FA, smo pri 150 osebkih iz obeh populacij določili število plavutnic v prsnih in trebušnih plavutih na obeh straneh telesa. Kot mero za asimetrijo smo uporabili absolutno razliko v številu plavutnic v prsnih in tre- bušnih plavutih, na vsaki strani telesa. Rezultati kažejo, da je za obe lastnosti FA večja pri diploidnih osebkih; razlika med populacijama je dovolj velika, da jo lahko štejemo kot statistič- no značilno le, če obe lastnosti obravnavamo združeno. Delo problematizira možnost presojanja sposobnosti preživetja le na osnovi FA in poudarja, da bi bilo nujno za določanje FA upora- biti metrične lastnosti. Ključne besede: ribe / kalifornijska postrv / Oncorhynchus mykiss / vitalnost / fluktuacijska asimetrija / razvojna stabilnost / sterilni triploidni organizmi / diploidne samice 1 INTRODUCTION Fluctuating asymmetry (FA) is the phenomenon observed in organisms which are bilaterally symmetrical. It occurs when the trait on one side of body differs in a random way from the same trait on the other side. It can be expressed as a difference of the trait observed on the left and right sides; the difference can be expressed as difference in number, size, shape or some other feature. FA is a measurement of developmental stability; devel- opmental homeostasis keeps FA at a low level. Increased level of FA reflects the fact that development of an organ- ism was instable. This instability could be caused either by genetic factors or by environment. The number of studies investigating the impact of miscellaneous envi- ronmental stressors on developmental stability is rather high and the variety of organisms studied is wide (Erik- sen et al., 2008). Two opposite hypotheses exist regard- Acta agriculturae Slovenica, 96/2 – 2010112 J. POHAR and K. STRGAR ing the relationship between genetic factors and FA; the first one claims that more heterozygous individuals have greater developmental stability which results in lower FA, while second one believes that when genes complexes which ensure developmental stability are disrupted (for instance by hybridization which increases heterozygot- ity) stability decreases and FA is elevated (Wilkins et al, 1995). In fish it is possible to experimentally manipulate the chromosome sets. By different techniques individuals of the same species with different sets of chromosomes originating from both parents or from one parent only could be produced (Komen and Thorgaard, 2007). In fish crossing of individuals from different species many times results in viable offspring. Fishes and specifically salmonids were therefore used to study the relation- ship between heterozygosity and developmental stability for types of heterozygotity introduced by manipulating chromosome sets or hybridization which could be found only exceptionally in natural populations (Leary et al., 1985; Wilkins et al., 1995; Young et al., 1995; Vøllestad and Hindar, 2001). Populations of rainbow trout where all individuals are either females (resulting from fertilizing ova of “nor- mal” females with sperm of individuals which are ge- netically females but produce sperm ) or sterile triploids (resulting from retention of second polar body in such ova induced by thermal or pressure shock) are widely used in aquaculture practice due to apparent advantage of such populations over populations of mixed sexes in production traits. In order to prevent reproduction of introduced species which could endanger native popula- tion it is also suggested that only such populations should be used for restocking waters where introduced species is not native (Aprehamian et al., 2001). The question of interest for aquaculturists and fish managers is whether sterile triploids exhibit larger viability than females. Ac- cording to Leary et al. (1984) viability of an organism and possibility to survive in natural environment could be judged by the magnitude of FA. The aim of our study was therefore to compare the difference in FA between two populations of rainbow trout; namely population of all-females and population of sterile triploids. 2 MATERIALS AND METHODS Number of pectoral and pelvic fin rays on left and right side of 150 females and 150 sterile triploids was de- termined by counting. After the fish were euthanized by electric shock the pectoral and pelvic fins on both sides were cut off and put for one minute into 50% solution of NaOH. The macerated skin and muscle tissue adjoining rays was removed by tweezers; the fin was put on blot- ting paper to dry up. The number of rays was established when pulled out from fin base one by one. All the count- ing was done by same person. For a batch of fish treated at the same day, the counting was consecutively done for each fish first on one side and then on the other side in order to avoid bias which could result from knowing the count on one side of individual fish while counting the number on the other side. For each fish and for each fin asymmetry was determined by subtracting the count on one side from the count on the opposite side. Both populations from which fish were sampled were of the same age (at the beginning of sampling 240 days after hatching) and reared under same condition at Fish research station Pšata which belongs to Biotechnical faculty of University of Ljubljana. Each day approximate- ly 20 fish from both groups were sampled and treated. Both groups were of the same genetic origin. Eggs stripped from broodstock kept at Fish research sta- tion Pšata were fertilized by sperm obtained from sex- inverted females using the methodology described by Ingram (1986). After fertilization half of eggs were put directly in incubation trays for incubation while half of eggs were firstly exposed to temperature shock and after that treated the same way as the eggs not exposed to tem- perature treatment. The heat shock was done according to slightly adapted method described by Ingram (1986). By heat shock triploid induction could range from 10 to 100% (Solar et al., 1984). Only those fish from second group were used for FA in which examination of gonads revealed sterility. 3 RESULTS The average, the lowest and the highest number of rays in pectoral and pelvic fins in both groups is shown in Table 1. The average number of pectoral rays is higher than average number of pelvic rays. The same was ob- served by Leary et al. (1985). The lowest number of pelvic and pectoral rays observed in diploids is lower than in triploids, while highest number does not differ between two groups. The average number of observed pectoral rays and pelvic rays found in our population is similar to numbers given by Leary et al. (1985). The difference in the average number of pelvic and pectoral rays between two groups is not statistically significant. The numbers obtained by subtracting the count of rays on one side of body from the count on the opposite side is shown in Table 2. These numbers are the meas- ure of asymmetry. If the count for specific trait was the same on both sides, such fish was not considered to be asymmetric for the trait under consideration. As it could Acta agriculturae Slovenica, 96/2 – 2010 113 FLUCTUATING ASYMMETRY IN DIPLOID FEMALE AND STERILE TRIPLOID RAINBOW TROUT (Oncorhynchus mykiss) be seen from the numbers in this table, fishes could ex- hibit the difference of 1, 2, 3 or 4; the latest number is the maximum difference in count for pectoral fin rays as well as pelvic fin rays. For both traits under consideration the majority of fish in both groups were not asymmetric; for pelvic fins around 80% of all fish did not exhibit asymme- try, for pectoral fins more than 60% of the fish were sym- metric. It is somehow surprising that pelvic fins which on average have lower number of rays are more asymmetric than pectoral fins. Figures indicate that the percentage of asymmetric individuals is higher in the population of diploids that in the population of triploids. The differ- ence between two populations is 4.7% for pectoral fins and 4.3% for pelvic fins. χ2 – test revealed that difference in numbers of individuals exhibiting 0, 1, 2, 3 or 4 dispar- ity in counts for each trait in two populations is not large enough to be considered statistically significant. On the basis of the same test done after pooling all individuals which were not symmetric into one group and comparing the numbers of animals in this group with the numbers of animals which were symmetric conclusion was alike even the test was done with lower degrees of freedom. For pectoral fins there were 84 sym- metric versus 66 asymmetric individuals in group of dip- loids and 91 symmetric versus 59 asymmetric individu- als in group of triploids. For pelvic fins the numbers of symmetric and asymmetric fish were 111 versus 39 for diploids and 119 versus 31 for triploids. The difference was too small to be statistically significant. Since we were measuring the magnitude of asym- metry (the number of units by which the left and right side differed) in two traits, we were able to summing the number of units of asymmetry for both traits for each fish. These numbers presented in Table 3 characterize the total magnitude of asymmetry for individual fish. By such methods the magnitude of asymmetry was increased; there were fishes for which the summed number of units by which the left and right side differed in both traits was as much as 6. χ2 – test revealed that difference in numbers of individuals exhibiting 0, 1, 2, 3, 4, 5 or 6 disparity in pooled counts for both traits in two populations is large enough to be considered statistically significant at P = 0.050. (calculated χ2 value was 12,762). The asymmetry measured by magnitude defined in such a way was increased; the numbers of symmetric and asymmetric fish were 65 versus 85 for diploids and 69 versus 81 for triploids. 4 CONCLUSIONS Our results demonstrate that sterile triploid rain- bow trout exhibit lower asymmetry than diploid females. The percentage of fish which were asymmetric was larger in the group of diploid females than in the group of ster- ile triploid for both traits investigated. Even the samples were rather large, these differences were not statistically significant when each trait was considered separately; the difference in pooled counts for both traits exhibited significance. Compared with the samples of similar ex- periment found in literature, samples used in our experi- ment were rather large. Wilkins et al. (1995) compared FA in diploids and triploids of Atlantic salmon (Salmo salar) and hybrids between Atlantic salmon and brown Number of rays Diploids Triploids Min Max Average Min Max Average Pectoral fins 24 32 28.56 26 32 28.81 Pelvic fins 16 23 20.94 19 23 21.04 Table 1: Average, minimum and maximum number of rays for pelvic and pectoral fins in diploid and triploid animals Preglednica 1: Povprečno ter najnižje in najvišje število pla- vutnic v prsnih in trebušnih plavutih diploidnih in triploidnih živali Asymmetry measured as L-R ray count (absolute value) Pectoral fins Pelvic fins Diploids Triploids Diploids Triploids Number of animals % Number of animals % Number of animals % Number of animals % 0 84 56.0 91 60.7 111 74.0 119 79.3 1 14 9.3 19 12.7 15 10.0 18 12.0 2 46 30.7 39 26.0 23 15.3 13 8.7 3 1 0.7 0 0.0 0 0.0 0 0.0 4 5 3.3 1 0.7 1 0.7 0 0.0 Total 150 100.0 150 100.0 150 100.0 150 100.0 Table 2: Number and percentage of diploid and triploid animals exhibiting (a)symmetry for pectoral and pelvic fins Preglednica 2: Število in odstotek diploidnih in triploidnih živali, ki kažejo (a)simetrijo v prsnih in trebušnih plavutih Acta agriculturae Slovenica, 96/2 – 2010114 J. POHAR and K. STRGAR trout (Salmo trutta). They collected data from 40 diploid salmon, 19 triploidised salmon, 41 hybrids and 41 trip- loidised hybrids. Their results indicate that triploidised salmon had FA values which were very similar to those of diploid salmon. There were 21% of triploids and 18% of diploids asymmetric in pectoral fin rays. The percent- age of asymmetric individuals in pelvic fins was 11% and 21% respectively. None of differences between two groups was statistically significant. In our experiment the percentage of asymmetric individuals for both traits un- der consideration was higher. The direct comparison of our results with these re- sults is not possible. Species examined by Wilkins et al. (1995) and species examined in our research belong two different groups of salmonids. Our results can also not to be directly compared to results of Leary et al. (1985) even they studied the same species as we did since the main goal of their research was to compare developmental sta- bility of gynogenetic diploid and triploid rainbow trout. Nevertheless, to some extend their results could be used for a comparison with ours as in addition to data col- lected on experimental population of triploids they also presented data on population of diploids collected in fish from commercial farms. These data show that triploids and diploids do not differ in absolute numbers of rays in pectoral and pelvic fins. For them this is one of the indicators that morphology of the triploid is similar to that of their diploid counterparts. The same can be con- cluded from our results shown in Table 1. The difference between diploids and triploids regarding these traits and the absolute numbers which were found for both traits in our research are similar to the results presented by them. Their conclusion about FA is the same as the one which can be done on the basis of our results: triploids had less FA than diploids. The mean magnitude of asymmetry which is the sum of the absolute values of the left minus right counts for all traits per individual were 1.67 and 1.76 for two strains of triploids compared to 2.08 and 1.90 for diploids of same strains. In our research these values were 0.97 for triploids and 1.29 for diploids. The differ- ence in absolute numbers resulted from the fact that in our research two meristic traits (pelvic and pectoral fins) were used to calculate this value while Leary et al. (1985) used five meristic counts. In research done on salmon and trout (Wilkins et al. 1995) values of mean magnitude of asymmetry were 0.90 for diploid and 0.79 for triploid salmon. Three meristic traits were used: number of gill rackers, number of rays in pelvic fins and number of rays in pectoral fins. In the focus of the question of interest for aquacul- turists and fish managers whether sterile triploids exhibit larger viability than females if they exhibit lower FA it is worth to mention the view of Leary et al. (1984) that lower FA in triploids does not indicate that triploidity has no deleterious effects on the development of rain- bow trout. Therefore the comparisons of viability of two populations should be done on the basis of other traits. The most appropriate would be to use traits which are important when fish is grown for human consumption or for restocking, like growth rate, survival, resistance to diseases. Majority of studies investigating FA in salmo- nids were based on meristic traits. (Young et al., 1995; Sánchez-Gálan et al. 1998; Young et al., 2009; Skog Erik- sen et al., 2008). The metric traits were used only in few studies. Vøllestad et al. (1998) for example measured the diameter of left and right eye as well as upper and lower left and right jaw length in grayling (Thymallus thymal- lus). General conclusion about methods used either for performing counts of meristic traits or measuring metric traits is that they are characterized as tedious tasks with low precision. Therefore it would be of great benefit for further research in this area to find method which could Units of asymmetry after summing L-R absolute difference for both traits Diploids Triploids Number of animals % Number of animals % 0 65 43.3 69 46.0 1 18 12.0 30 20.0 2 45 30.0 43 28.7 3 9 6.0 3 2.0 4 9 6.0 5 3.3 5 1 0.7 0 0.0 6 3 2.0 0 0.0 Total 150 100.0 150 100.0 Table 3: Number and percentage of diploid and triploid animals exhibiting (a)symmetry for pectoral and pelvic fins Preglednica 3: Število in odstotek diploidnih in triploidnih živali, ki kažejo (a)simetrijo v prsnih in trebušnih plavutih Acta agriculturae Slovenica, 96/2 – 2010 115 FLUCTUATING ASYMMETRY IN DIPLOID FEMALE AND STERILE TRIPLOID RAINBOW TROUT (Oncorhynchus mykiss) be used to count or measure traits with higher precision and speed. 5 REFERENCES Aprahamian M. W., Martin Smith K., McGinnity P., McKelvey S., Taylor J. 2003. Restocking of salmonids – opportunities and limitations. Fisheries Research, 62, 2: 211–227 Leary R.F., Allendorf F.W., Knudsen K.L. 1984. Superior devel- opmental stability of heterozygotes at enzyme loci in sal- monid fishes. Am. Nat., 124: 540–551 Leary R.F., Allendorf F.W., Knudsen K.L., Thorgaard G.H. 1985. Heterozygosity and developmental stability in gynogenetic diploid and triploid rainbow trout. Heredity, 54: 219–225 Eriksen M.S., Espmark Å.M., Poppe T., Braastad B.O., Salte R., Bakken M. 2008. Fluctuating asymmetry in farmed Atlan- tic salmon (Salmo salar) juveniles: also a maternal matter? Environ. Biol. Fish., 81: 87–99 Ingram M. 1986. Clearwater guide to investment in ova and milt technology broodstock management. Surby, Clearwa- ter Publishing Limited: 73 p. Komen H., Thorgaard G.H. 2007. Androgenesis, gynogenesis and the production of clones in fishes: A review. Aquacul- ture, 269, 1–4: 150–173 Sánchez-Galán S., Linde A.R., Izquierdo J.I., García-Vázquez E. 1998. Micronuclei and fluctuating asymmetry in brown trout (Salmo trutta): complementary methods to biomoni- tor freshwater ecosystems. Mutation Research/Genetic Toxicology and Environmental Mutagenesis, 412, 3: 219– 225 Solar I.I., Donaldson E.M., Hunter G.A. 1984. Induction of triploidy in rainbow trout (Salmo gairdneri Richardson) by heat shock, and investigation of early growth. Aquaculture, 42, 1: 57–67 Vøllestad L.A., Fjeld E., Haugen T., Oxnevad S.A. 1998. Devel- opmental instability in grayling (Thymallus thymallus) ex- posed to methylmercury during embryogenesis. Environ- mental Pollution, 101, 3: 349–354 Vøllestad L.A., Hindar K. 2001. Developmental stability in brown trout: are there any effects of heterozygosity or envi- ronmental stress? Biological Journal of the Linnean Society, 74, 3: 351–364 Wilkins N.P., Gosling E., Curatolo A., Linnane A., Jordan C., Courtney H.P. 1995. Fluctuating asymmetry in Atlantic salmon, European trout and their hybrids, including trip- loids. Aquaculture, 137, 1–4: 77–85 Young W.P., Wheeler P.A., Thorgaard G.H. 1995. Asymmetry and variability of meristic characters and spotting in iso- genic lines of rainbow trout. Aquaculture, 137, 1–4: 67–76 Young W.P., Frenyea K., Wheeler P.A., Thorgaard G.H. 2009. No increase in developmental deformities or fluctuating asym- metry in rainbow trout (Oncorhynchus mykiss) produced with cryopreserved sperm. Aquaculture, 289, 1–2: 13–18 Acta argiculturae Slovenica, 96/2, 117–125, Ljubljana 2010 COBISS: 1.01 Agris category code: L10 OCENA PARAMETROV DISPERZIJE ZA LASTNOSTI ZUNANJOSTI PRI KONJIH HAFLINŠKE PASME Martina PLANINC 1, Janez RUS 2, Milena KOVAČ 3, Špela MALOVRH 4 Delo je prispelo 20. oktobra 2010, sprejeto 26. novembra 2010. Received October 20, 2010; accepted November 26, 2010. 1 Univ. v Ljubljani, Biotehniška fak., Odd. za zootehniko, Groblje 3, SI-1230 Domžale, Sovenija, assist., e-mail: martina.planinc@bf.uni-lj.si 2 Univ. v Ljubljani, Veterinarska fak., Inštitut za rejo in zdravstveno varstvo kopitarjev, C v Mestni log 47, SI-1000 Ljubljana, Sovenija, mag., e-mail: janez.rus@vf.uni-lj.si 3 Isti naslov kot 1, prof. dr., e-mail: milena.kovac@bf.uni-lj.si 4 Isti naslov kot 1, znan. sod., e-mail: spela.malovrh@bf.uni-lj.si Ocena parametrov disperzije za lastnosti zunanjosti pri konjih haflinške pasme V naši raziskavi smo pri konjih haflinške pasme v Slove- niji za lastnosti zunanjosti ocenjevali komponente (ko)varianc. V podatkovni zbirki je bilo skupaj 3371 živali, od tega smo jih v raziskavo vključili 600 (15 žrebcev in 585 kobil). Živali, ki smo jih vključili v analizo, so imele zapise desetih ocen in/ali deve- tih meritev ter znanega vsaj enega od staršev. Model je za vse ocenjene in izmerjene lastnosti vseboval leto ocenjevanja in/ali merjenja kot sistematski vpliv in naključni vpliv živali. Upora- bili smo metodo omejene največje zanesljivosti (REML) v pro- gramu VCE. Pozitivno definitne matrike smo dobili s pomočjo postopka, ki se imenuje ukrivljanje matrik (ang. bending). Za ocenjene lastnosti so znašale heritabilitete od 0,40 za prednji del trupa do 0,78 za pasemsko značilnost. Heritabilitete za izmerje- ne lastnosti so se gibale med 0,20 za globino prsi in 0,62 za viši- no vihra, merjeno s palico. Genetske korelacije so bile v večini pozitivne. Najvišja genetska korelacija pri ocenjenih lastnostih je 0,92 med skupno oceno in zadnjim delom trupa. Med oceno pasemske značilnosti in oceno prednjih nog korelacije ni bilo. Pri izmerjenih lastnostih so bile genetske korelacije ocenjene od 0,38 med dolžino trupa in obsegom prsi do 0,95 med višino vihra, merjeno s palico in višino vihra, merjeno s trakom. Ključne besede: konji / pasme / haflinška pasma / ha- flinger / lastnosti zunanjosti / selekcija / genetski parametri / Slovenija Estimation of dispersion parameters for linear type traits in the Haflinger horses The covariance components for exterior traits were esti- mated on Haflinger horses in Slovenia. There were 3371 data in- cluded in the database. Data from 600 animals (15 stallions and 585 mares) with known pedigree were analysed. For each horse, at most ten traits were scored and nine traits were measured. The fixed part of the model included only the year when horse was scored or measured and animal was treated as random ef- fect. Genetic and environmental parameters for exterior traits were estimated by the restricted maximum likelihood method (REML) as implemented in the program package VCE. To make matrices positive definite we used a statistic method commonly known as ’bending’. Heritabilities for the scored traits were es- timated between 0.40 for front body part and 0.78 for the breed type. For measured traits the heritabilities were between 0.20 for chest depth and 0.62 for withers height (measuring stick). Genetic corelations were in most cases positive. The highest ge- netic corelation for scored traits was 0.92 between total score and rear body part. There was no corelation betwen breed type and front legs. Genetic corelations for measured traits were from 0.38 between body lenght and chest size to 0.95 betwen withers hight measured with stick and measured with tape. Key words: horses / breeds / Haflinger / exterior traits / selection / genetic parameters / Slovenia Acta agriculturae Slovenica, 96/2 – 2010118 M. PLANINC in sod. 1 UVOD Haflinška pasma konj je nastala na Južnem Tirol- skem in je razširjena v več kot petdesetih državah po vseh kontinentih (Viliani, 2008). Konj haflinške pasme je majhnega okvirja z značilno barvo lisjaka in s plavo grivo in repom. Povprečna višina vihra konj haflinške pasme je med 140 in 155 cm. Kobile so nekoliko nižje od žrebcev. Telesna masa pri odrasli živali je okoli 500 kg. Konjev ha- finške pasme ne uvrščamo niti med toplokrvne niti med hladnokrvne. V Sloveniji se ta pasma vzreja kot tradicio- nalna pasma konj (Pravilnik o ohranjanju ..., 2004) in se razvija v tipu ljubiteljskega, jahalnega konja (Rus, 2005). Odbira živali poteka na podlagi desetih ocen in/ali deve- tih meritev, ki so opisane v rejskem programu za konje haflinške pasme (Rus, 2005). Napovedovanje plemenske vrednosti konj poteka na osnovi ocen in meritev, ki se opravljajo direktno na živali (Arnason in Van Vleck, 2000). Pri izboru lastnosti, ki so pomembne za selekcijo, je potrebno upoštevati dednost, merljivost, gospodarski in biološki pomen lastnosti. Pri različnih pasmah konj po svetu se plemenske vrednosti napoveduje predvsem za delovne in tekmovalne sposob- nosti (Langlois in Blouin, 2004). Pri tem se vključuje po- datke z ocenjevanja in merjenja konj ter njihovo poreklo. Za napovedovanje plemenske vrednosti v konjere- ji najbolj razširjena metoda mešanih modelov. Metoda omogoča hkratno ocenjevanje genetskih in okoljskih vplivov in je postala standardna metoda za genetsko vre- dnotenje konj (Arnason, 1996). V tujini pri sestavljanju modela pri različnih populacijah upoštevajo veliko raz- ličnih vplivov. Kot vplive v modelih uporabljajo spol, sta- rost, rejca, popisovalca in ocenjevalca, geografsko regijo, genetsko skupino in kondicijo (van Bergen Henk in van Arendonk, 1993; Koenen in sod., 1995; Dolvik in Klemet- sdal, 1999). V Italiji s to metodo napovedujejo plemensko vrednost konjem haflinške pasme za linearne lastnosti te- lesnega ustroja (Samoré in sod., 1997), na Islandiji vsem konjem, ki so vključeni v podatkovno zbirko (Huganson, 1994), na Poljskem konjem arabske pasme (Sobczyin- ska in Kownacky, 1996), kasačem v Franciji (Langlois in Vrijenhoek, 2004) ter kasačem na Norveškem (Arnason, 1996). Parametre disperzije za ocenjene lastnosti zuna- njosti so ocenili pri konjih haflinške pasme v Italiji (Sa- moré in sod., 1997), za izmerjene lastnosti pri haflinških konjih pa bomo rezultate predstavili prvi, saj jih za tuje populacije haflinških konj v literaturi nismo zasledili. Za ocenjevanje lastnosti zunanjosti se v Sloveni- ji uporablja skala z opisnimi ocenami od 1 do 10 brez vmesnih ocen (Rus, 2005). Ocena 0 pomeni, da žival ni bila ocenjena za to lastnost in ima manjkajočo vrednost. Posamezne ocene za lastnosti zunanjosti, ki predstavlja- jo število točk, se seštejejo v skupno oceno in na podla- gi skupne ocene se živali razvrščajo v šest kakovostnih razredov (1a, 1b, 2a, 2b, 3a, in 3b). Znotraj razredov se pri mejnih vrednostih pripiše oznaka plus (+) pri zgor- nji meji ali minus (−) pri spodnji meji. V Italiji (Samoré in sod., 1997) linearno ocenjujejo pri konjih haflinške pasme 10 sklopov lastnosti, znotraj katerih je skupaj 26 posameznih lastnosti ocenjenih z ocenami od 0 do 10 na pol točke natančno. Pri andaluzijskih konjih v Španiji (Molina in sod., 1999) linearno ocenjujejo 11 lastnosti, ki so razporejene v dveh sklopih: morfološke lastnosti in ocene pasemskega tipa. Pri šetlandskih ponijih (van Ber- gen Henk in van Arendonk, 1993) smo lahko zasledili 28 ocen lastnosti zunanjosti in pri nizozemskem toplokrv- nem konju 26 lastnosti, ki so ocenjene na linearni točk- ovni skali z vrednostimi od 0 do 40. Pri konjih haflinške pasme v Sloveniji merimo naj- več osem lastnosti zunanjosti (Rus, 2005). V Braziliji so na konjih panteniro pasme izvajajo 14 meritev (Miserani in sod., 2002) in na nizozemskih vlečnih konjih kar 31 meritev (Druml in sod., 2008). V Sloveniji je minimal- na višina vihra, merjena s palico, za vpis žrebcev v elitno knjigo 142 cm in 140 cm za vpis žrebcev v knjigo žrebcev II in kobil v glavno knjigo kobil. Višina 138 cm je mi- nimalna pri vpisu kobil v splošno rodovniško knjigo in kobile morajo dosegati vsaj 136 cm višine vihra, merjeno s palico, za vpis v evidenčno knjigo. Pri odbiri konj je pomembna tudi informacija o oceni delovne sposobnosti konj (Rus, 1996). Preizkus delovnih sposobnosti haflingerjev je v tujini določen z rejskim programom. Preizkus delovnih sposobnosti za toplokrvne konje v Sloveniji se ni opravljali do leta 1996, predvsem zaradi majhnega števila konj in slabe kakovo- sti plemenskega materiala (Rus, 1996). Po letu 1996 se je povečal interes za rejo toplokrvnih konj. Kakovost ple- menskega materiala se je izboljšala in preizkus delovnih sposobnosti je tudi pri nas postal pomemben vir infor- macij za napoved plemenske vrednosti. Namen našega dela je bil proučiti lastnosti zunanjo- sti pri konjih haflinške pasme in postaviti večlastnostni statistični model, primeren za napovedovanje plemenske vrednosti v Sloveniji. 2 MATERIAL IN METODE 2.1 MATERIAL Podatke o konjih haflinške pasme v Sloveniji smo dobili z Inštituta za zdravstveno varstvo kopitarjev na Veterinarski fakulteti, kjer opravljajo identifikacijo, regi- stracijo in vpis živali v rodovniško knjigo. Skupno šte- vilo živali v podatkovni zbirki je bilo 3371. V statistične analize za oceno parametrov disperzije so bili vključeni Acta agriculturae Slovenica, 96/2 – 2010 119 OCENA PARAMETROV DISPERZIJE ZA LASTNOSTI ZUNANJOSTI PRI KONJIH HAFLINŠKE PASME podatki konj haflinške pasme, ki so bili vneseni in urejeni v podatkovni zbirki do marca 2008. Podatke iz podatkovne zbirke smo uredili s pro- gramskim jezikom SQL, pripravili smo datoteko z oce- njenimi in izmerjenimi lastnostmi ter datoteko s pore- klom živali. Iz podanih posameznih ocen o pasemskih značilnosti, vratu, glavi, prednjem, srednjem in zadnjem delu trupa, prednjih in zadnjih nogah ter pravilnosti in izdatnosti hodov smo izračunali skupno oceno za živali, ki so imele podane vseh deset ocen. Na živalih so bile opravljene naslednje meritve: višina vihra, merjena s trakom in palico, višina križa, merjena s trakom, obseg prsi in piščali, širina prsi in križa, globina prsi in dolžina trupa. V obdobju od leta 1990 do marca 2008 je tako bilo ocenjenih 600 živali. Število ocenjenih živali se je med leti spreminjalo. V povprečju je bilo na leto ocenjenih in/ ali merjenih 31,6 živali. Največ (64) jih je bilo ocenjenih v letu 2005. Starost živali je bila od dveh pa do 19 let. Kar 68,5 % živali je bilo ocenjenih in/ali merjenih pri starosti treh let (slika 1). Živali, ki so bile stare med pet in vključ- no devet let, so predstavljale 9,6 % vseh ocenjenih in/ali merjenih živali. Kar 1,7 % živali je bilo starih deset let ali več. V letu 2008 je bilo do marca ocenjenih in/ali merje- nih 13 živali, starih med sedem in deset let. Pripravljeno poreklo je zajemalo skupaj 1956 živali z globino osem generacij (pregl. 1). Prvo generacijo so predstavljale živali, ki so imele vsaj pet ocen in/ali me- ritev. V predhodne generacije so bili uvrščeni njihovi predniki. V poreklu je imelo 80,57 % živali znanega vsaj enega starša, oba starša sta bila znana pri 79,47 % živalih. 0 10 20 30 40 50 60 70 80 2 3 4 5 6 7 8 9 10 11 12 14 15 19 Starost (leta) D el ež ž iv al i ( % ) . Slika 1: Porazdelitev živali glede na starost ob ocenjevanju in/ali merjenju. Figure 1: The distribution of animals according to age at evaluation and / or measurement. Generacija Št. živali Oba starša znana Znan oče Znana vsaj en starš Neznani starši 1 600 559 1 600 - 2 280 280 - 280 - 3 227 227 - 227 - 4 294 270 5 275 19 5 338 148 5 153 185 6 173 25 6 31 142 7 35 6 2 8 27 8 9 1 1 2 7 Skupaj 1956 1556 20 1576 377 Delež (%) 199 79,55 1,02 80,57 19,27 Preglednica 1: Struktura porekla Table 1: The origin structure Acta agriculturae Slovenica, 96/2 – 2010120 M. PLANINC in sod. Osnovno oziroma izhodiščno populacijo so predstavlja- le živali z neznanimi starši (377 oziroma 19,27 %). Med konji haflinške pasme, ki smo jih vključili v analizo, je bilo 71 različnih očetov in 363 mater. V povprečju je bilo na vsakega očeta skoraj 28 potomcev in na vsako mater 5 potomcev. 2.2 METODE Podatkovno zbirko za kopitarje v Sloveniji vzpo- stavljamo iz starih zapisov. Nabor možnih vplivov, ki bi jih vključili v model, je bil v primerjavi s tujo literaturo manjši. Pri izboru končnega modela smo upoštevali sta- tistično značilnost vplivov (p-vrednost), koeficient deter- minacije (R2) in število stopinj prostosti za posamezne vplive in za model v celoti. Razlike med spoloma niso bile statistično značilne, kar je verjetno posledica neu- godne strukture podatkov po spolu. Žrebci so bili pre- malo zastopani, ocenjevalna skala pa že upošteva spolni dimorfizem. Starost, ki je predstavljala razliko med letom ocenjevanja in letom rojstva, se je pri nekaterih lastnostih pokazala za značilno, vendar pa bi vključitev vpliva v mo- del le malo doprinesla k pojasnjeni varianci. Po predhodnih statističnih analizah smo uporabili enostavni model (enačba 1), ki je vključeval samo leto ocenjevanja in/ali merjenja (Li) kot sistematski vpliv. Na- ključni vpliv je predstavljala žival oziroma aditivni genet- ski vpliv (aij). yij = μ + Li + aij + eij (1) Pri napovedovanju plemenskih vrednosti se lahko uporabi samo pozitivno definitne matrike genetskih va- rianc in kovarianc (Hayes in Hill, 1981). Pozitivna defini- tnost matrik je bil zato naš kriterij za izbor lastnosti zu- nanjosti pri konjih haflinške pasme. Pozitivno definitnost matrik smo preverjali s Cholesky razčlenitvijo (Hayes in Hill, 1981; Jorjani in sod., 2003). Ker matrika merjenih lastnosti ni bila pozitivno definitna, smo naredili ukri- vljanje matrike (bending). Razvoj sistematskega dela modela smo opravili po metodi najmanjših kvadratov s proceduro GLM v sta- tističnem paketu SAS (SAS Inst. Inc., 2001). Parametre disperzije smo analizirali z metodo omejene največje zanesljivosti (REML) v programu VCE (Kovač in sod., 2002). Podatke smo za ta program predhodno pripravili s programom PEST (Groeneveld in sod., 1990). 3 REZULTATI IN RAZPRAVA V genetsko analizo smo vključili 600 živali, ki so imele podatke o ocenjevanju in merjenju. Rezultate smo razdelili v dva sklopa. V prvem delu predstavljamo feno- tipske vrednosti in v drugem koeficiente determinacije. 3.1 FENOTIPSKE VREDNOSTI ZA LASTNOSTI ZUNANJOSTI Pri ocenjenih lastnostih zunanjosti je bilo največ ocenitev (600) opravljenih za pasemsko značilnost, glavo in vrat ter najmanj ocen (561) za izdatnost hodov (pre- gl. 2). Najvišje povprečje je bilo pri ocenah za prednji del trupa, kjer je bila povprečna ocena 7,6 in najmanjše za lastnost zadnjih nog, kjer je bila povprečna ocena 6,7. Povprečje skupne ocene je bilo dobrih 71 točk. Pri konjih haflinške pasme v Italiji so se povprečne vrednosti ocen gibale med 4,5 in 5,5 (Samoré in sod., 1997), kar bi lahko Št. ocen Povprečje Standardni odklon Modus Minimum Maksimum Skupna ocena 561 71,4 4,72 69 58 87 Pasemska značilnost 600 7,3 0,70 7 6 10 Glava 600 7,2 0,82 7 5 9 Vrat 600 7,3 0,77 7 6 9 Prednji del trupa 599 7,6 0,68 8 5 9 Srednji del trupa 599 7,1 0,74 7 5 9 Zadnji del trupa 599 7,3 0,67 7 5 9 Prednje noge 598 6,8 0,92 7 4 9 Zadnje noge 598 6,7 0,73 7 4 9 Pravilnost hodov 574 6,9 0,80 7 4 9 Izdatnost hodov 561 7,4 0,77 7 5 10 Preglednica 2: Opisna statistika ocen za lastnosti zunanjosti Table 2: Descriptive statistics for estimated traits Acta agriculturae Slovenica, 96/2 – 2010 121 OCENA PARAMETROV DISPERZIJE ZA LASTNOSTI ZUNANJOSTI PRI KONJIH HAFLINŠKE PASME pripisali večjemu številu ocenjenih živali in uporabi celo- tne skale za ocenjevanje. Med posameznimi lastnostmi je največji standardni odklon v našem primeru imela ocena prednjih nog (0,9) in najmanjši standardni odklon ocena prednjega dela trupa (0,7). Standardni odkloni so bili v raziskavi, ki so jo opravili Samoré in sod. (1997), med 0,9 za poslušnost in 2,0 za barvo grive. Standardni odkloni v našem primeru pri nobeni posamezni lastnosti ne dose- žejo pričakovane vrednosti, ki znaša 1,66. Modus za vse ocenjene lastnosti, z izjemo ocen za prednji del trupa, je znašal 7 (pregl. 2). Razpon ocen na ocenjevalni skali je od 1 do 10, ven- dar v praksi vidimo, da je bila najmanjša podeljena ocena 4 (pregl. 2) pri oceni nog in hodov. Od vseh 5928 dode- ljenih ocen v naši raziskavi je bila najbolj pogosta ocena 7, uporabljena 2921-krat, kar predstavlja polovico vseh podeljenih ocen. Razpon med ocenami za lastnosti je bil največ šest točk. Na podlagi skupne ocene se konji razvrstijo v kako- vostne razrede. Nobena žival v našem poskusu ni prese- gla 87 točk, zato je 1a razred v našem primeru prazen, saj je zanj potrebnih 90 točk in več (slika 2). Večina živali je bila razvrščena med +3a in +2b kakovostnim razredom. V najslabši razred 3b je bilo razvrščenih 45 živali (7,5 %). Podatki, ki smo jih obdelali, nakazujejo možnost, da je bila prva odbira opravljena pred ocenjevanjem. Odbiro opravijo že sami rejci, ki konje s slabšimi lastnostmi ne pripeljejo na ocenjevanje. Porazdelitve ocen za posame- 0 5 10 15 20 25 30 +1b 1b -1b +2a 2a -2a +2b 2b -2b +3a 3a 3b Razredi D el ež ž iv al i ( % ) . Kobile Žrebci Slika 2: Porazdelitev živali v kakovostne razrede ločeno po spolu. Figure 2: Animals in classes by gender. 0 10 20 30 40 50 60 4 5 6 7 8 9 Ocene D el ež (% ) Pravilnost hodov Prednji del trupa Slika 3: Porazdelitev ocen za lastnosti pravilnost hodov in prednji del trupa. Figure 3: Distribution of accuracy of walk and the front of body. Acta agriculturae Slovenica, 96/2 – 2010122 M. PLANINC in sod. zne lastnosti naj bi bile simetrične. Ocene za pravilnost hodov so v primerjavi z ostalimi lastnostmi porazdeljene dokaj simetrično (slika 3). Zajete so ocene od 4 do 10, najpogostejša vrednost pa je 7. Za primerjavo prikazuje- mo še porazdelitev ocen za prednji del trupa, kjer so bile uporabljene le štiri ocene na intervalu od 5 do 9 z najpo- gostejšo oceno 8. Opisna skala je naravnana tako, da olaj- ša odbiro, ni pa najbolj primerna za obdelavo podatkov. Pri konjih haflinške pasme v Sloveniji (pregl. 3) je bilo največ meritev (593) opravljenih za višino vihra, merjeno z merilno palico, ter obseg prsi in najmanj (450) za dolžino trupa in širino križa. V povprečju je bila višina vihra, merjena s palico, 139,3 cm s standardnim odklo- nom 3,2 cm. Meritev vihra s trakom je bila za 10,9 cm višja. V povprečju bi se lahko vse živali vpisale v splošno rodovniško knjigo, kjer je pogoj minimalna višina vihra merjena s palico 138 cm. Živali, ki so imele višino vihra nižjo od 136 cm, niso zadostile pogojem za vpis v evi- denčno knjigo. Višina križa je bila v povprečju za 1,33 cm višja kot višina vihra, merjena s palico. Obseg piščali je bil v povprečju 19,31 cm in obseg prsi 177,24 cm, kjer je bil tudi največji standardni od- klon, ki je znašal 8,56 cm. Variabilnost merjenih živali je bila majhna. Kadar je majhna variabilnost posledica predhodne odbire, otežuje selekcijsko delo in zmanjšuje učinkovitost rejskega dela. Št. meritev Povprečje Standardni odklon Minimum Maksimum Višina vihra – palica (cm) 593 139,34 3,21 132 152 Višina vihra – trak (cm) 592 150,20 4,27 140 167 Obseg prsi (cm) 593 177,24 8,56 152 218 Globina prsi (cm) 451 65,29 2,74 50 72 Širina prsi (cm) 454 41,96 3,67 30 64 Obseg piščali (cm) 586 19,31 0,93 17 22 Višina križa (cm) 455 140,67 4,71 131 151 Širina križa (cm) 450 52,75 3,23 34 88 Dolžina trupa (cm) 450 150,82 5,41 135 169 Preglednica 3: Opisna statistika za izmerjene lastnosti Table 3: Descriptive statistics for measured traits S PZ GL VR PDT SDT ZDT PN ZN PH IH Fenotip. var. S 0,64 1,55 −0,10 1,53 1,33 1,07 2,26 1,86 1,74 2,01 2,23 24,27 PZ 0,53 0,68 −0,12 0,24 0,14 0,11 0,30 0,00 0,11 0,05 0,16 0,82 GL −0,03 −0,21 0,78 −0,09 −0,05 0,12 −0,03 −0,09 −0,16 −0,15 −0,11 0,73 VR 0,65 0,51 −0,18 0,52 0,23 0,21 0,20 −0,04 0,13 0,07 0,20 0,76 PDT 0,81 0,45 −0,15 0,87 0,40 0,13 0,17 0,08 0,13 0,14 0,20 0,44 SDT 0,42 0,22 0,25 0,52 0,49 0,61 0,06 −0,09 −0,05 −0,02 0,17 0,68 ZDT 0,92 0,65 −0,07 0,51 0,65 0,16 0,53 0,31 0,29 0,29 0,27 0,73 PN 0,63 0,00 −0,16 −0,08 0,24 −0,18 0,66 0,52 0,39 0,47 0,25 1,08 ZN 0,67 0,22 −0,32 0,29 0,47 −0,11 0,72 0,79 0,65 0,34 0,14 0,66 PH 0,75 0,11 −0,29 0,16 0,78 −0,04 0,69 0,91 0,77 0,62 0,34 0,74 IH 0,74 0,28 −0,19 0,42 0,63 0,36 0,57 0,44 0,27 0,65 0,53 1,09 Preglednica 4: Ocena heritabilitet (diagonala), kovarianc (nad diagonalo) za aditivni genetski vpliv , genetske korelacije (pod diago- nalo) ter ocena fenotipskih varianc (zadnji stolpec) za ocenjene lastnosti Table 4: Estimates of heritability (diagonal) covariance (above diagonal) for additive genetic effects, genetic correlations (below diago- nal) and phenotypic variance estimate (last column) for the evaluated traits S – skupna ocena, PZ – pasemska značilnost, GL – glava, VR – vrat, PDT – prednji del trupa, SDT – srednji del trupa, ZDT – zadnji del trupa, PN – prednje noge, ZN – zadnje noge, PH – pravilnost hodov, IH – izdatnost hodov Acta agriculturae Slovenica, 96/2 – 2010 123 OCENA PARAMETROV DISPERZIJE ZA LASTNOSTI ZUNANJOSTI PRI KONJIH HAFLINŠKE PASME 3.2 OCENE (KO)VARIANC ZA LA STNOSTI ZUNA- NJOSTI 3.2.1 OCENJENE LASTNOSTI V analizo parametrov disperzije smo zajeli vse oce- njene lastnosti. Fenotipske variance so se gibale med 0,44 in 1,09 z izjemo skupne ocene, kjer je fenotipska vari- anca 24,27 (pregl. 4). Genetska varianca je bila pričako- vano največja (15,63) pri seštevku ocen. Pri posameznih lastnostih je najmanjša genetska varianca (0,14) bila pri oceni za srednji del trupa. Aditivne genetske variance so Albertsdóttir in sod. (2007) v raziskavi, kjer so ocenjeva- li genetske parametre za lastnosti zunanjosti in rezultate tekmovanj pri islandskih konjih, ocenili med 0,06 in 0,94. Od primerljivih lastnosti imajo najvišje (0,11) ocenjeno varianco za aditivni genetski vpliv lastnosti kakovosti nog (v našem poskusu je bila ocenjena nad 0,40) in naj- nižje (0,06) ocenjeno stojo nog. Aditivno genetsko vari- anco za glavo in vrat so ocenili na 0,07 ter za zadnji del trupa 0,08, kar je precej manj, kot je bilo v našem prime- ru. Razlike v ocenah genetskih varianc so lahko posledica razlik v pasmi in različnih ocenjevalnih skal. Heritabilitete pri haflinških konjih v Sloveniji so se gibale med 0,40 za prednji del trupa in 0,78 za oceno glave (pregl. 4). V primerjavi z našimi rezultati so nižje heritabilitete za lastnosti zunanjosti ocenili pri konjih ha- flinške pasme v Italiji (Samoré in sod., 1997). Za lastnost vratu so heritabiliteto izvrednotili na 0,04 in za lastnosti nog med 0,10 in 0,17. Pri nizozemskih toplokrvnih ko- njih (Koenen in sod., 1995), kjer je skala z ocenami od 0 do 40, so bile heritabilitete za ocenjene lastnosti med 0,09 in 0,28. Pri šetlandskih ponijih (van Bergen Henk in van Arendonk, 1993) so heritabiliteto za izdatnost hoda oce- nili na 0,35 (pri nas 0,53) in za zadnje noge 0,07 (pri nas 0,65). Pri islandskih konjih (Albertsdóttir in sod., 2007) so za lastnosti glave, nog in hoje podali nekoliko nižje re- zultate. Heritabiliteto za prednji del trupa pri populaciji vlečnih konj na Nizozemskem so Druml in sod. (2008) ocenili na 0,16, v našem primeru smo jo na 0,40. Suonta- ma in sod. (2009) so heritabiliteto za lastnosti nog ocenili na 0,13 in za lastnosti gibanja 0,17. Genetske korelacije so bile v večini pozitivne. Ne- gativne korelacije smo izračunali med oceno za glavo in ostalimi lastnostmi. Genetska korelacija med prednjimi in zadnjimi nogami je ocenjena na kar 0,79. Genetsko ko- relacijo med prednjimi in zadnjimi nogami so Samoré in sod. (1997) ocenili na 0,06. Najvišjo genetsko korelacijo smo ocenili na 0,92 med skupno oceno in zadnjim delom trupa. Samoré in sod. (1997) so močne genetske kore- lacije ocenili med lastnostmi vratu (0,94) in med posa- meznimi lastnostmi glave (0,99). Višje ocenjene genetske korelacije so bile tudi pri nizozemskih konjih (Koenen in sod., 1995). Šibko genetsko korelacijo (0,07) so Druml in sod. (2008) ocenili med glavo in pravilnostjo hodov. Do razlik med ocenami parametrov disperzije lahko prihaja zaradi različnih pasem, zaradi postopka ocenjevanja in ocenjevalne skale. Največjo težavo pri primerjavi pred- stavlja opisna skala, ki se uporablja pri ocenjevanju konj v Sloveniji. V tujini se za ocenjevanje konj uporabljajo linearne skale. 3.2.2 PARAMETRI DISPERZIJE ZA IZMERJENE LA- STNOSTI Fenotipske variance so bile med 18,06 cm2 za višino vihra, merjeno s palico, in 0,76 cm2 za obseg piščali (pre- VVT VVP GP ŠP OP OPI VK ŠK DT Fenotip. var. VVT 0,57 7,17 6,39 2,67 2,63 0,80 4,92 1,86 5,44 8,90 VVP 0,95 0,62 10,50 3,55 4,45 1,10 7,54 3,59 8,37 18,06 GP 0,73 0,81 0,20 4,48 5,08 1,23 6,69 5,38 8,64 7,70 ŠP 0,68 0,61 0,66 0,43 1,50 0,50 1,98 1,39 3,39 7,08 OP 0,62 0,70 0,69 0,46 0,29 0,39 2,42 2,41 2,31 12,55 OPI 0,72 0,67 0,65 0,59 0,42 0,32 0,76 0,46 1,22 0,76 VK 0,92 0,94 0,72 0,48 0,54 0,65 0,51 2,08 5,92 11,09 ŠK 0,53 0,68 0,89 0,51 0,81 0,59 0,56 0,24 3,04 10,38 DT 0,74 0,77 0,68 0,60 0,38 0,76 0,76 0,60 0,36 11,99 Preglednica 5: Ocena heritabilitet (diagonala), kovarianc (nad diagonalo) za aditivni genetski vpliv , genetske korelacije (pod diago- nalo) ter ocena fenotipskih varianc (zadnji stolpec) za izmerjene lastnosti Table 5: Estimates of heritability (diagonal) covariance (above diagonal) for additive genetic effects, genetic correlations (below diago- nal) and phenotypic variance estimate (last column) for the measured traits VVT – višina vihra, merjena s trakom, VVP – višina vihra, merjena s palico, GP – globina prsi, ŠP – širina prsi, OP – obseg prsi, OPI – obseg piščali, VK – višina križa, ŠK – širina križa, DT – dolžina trupa Acta agriculturae Slovenica, 96/2 – 2010124 M. PLANINC in sod. gl. 5). Genetske variance pri izmerjenih lastnostih so bile ocenjene dokaj majhne (pregl. 5). Genetska varianca za višino vihra, merjeno s palico, je bila ocenjena na 11,23 cm2, za dolžino trupa 10,58 cm2 in za globino prsi na 15,08 cm2. Najnižja genetska varianca je bila pri obsegu piščali, kjer je bila ocenjena na 0,24 cm2. Genetske vari- ance za merjene lastnosti so ocenjevali na nizozemskih toplokrvnih konjih (Dolvik in Klemetsdal, 1999). Za viši- no vihra, merjeno s trakom, sta variance ocenila na 12,25 cm2 in za obseg piščali 0,38 cm2. Za izmerjene lastnosti so ocene heritabilitet prese- gale 0,20 (pregl. 5). Heritabilitete so bile visoke za višino vihra, merjeno s palico (0,62) ali trakom (0,57) ter višino vihra (0,51). Najnižjo (0,20) heritabiliteto je imela v na- šem primeru globina prsi. Pri norveških hladnokrvnih konjih (Dolvik in Klemetsdal, 1999) je bila heritabiliteta za višino vihra, merjeno s palico, ocenjena na 0,73, pri islandskih konjih na 0,67 (Albertsdóttir in sod., 2007) in pri andaluzijskih konjih v Španiji na 0,50 (Molina in sod., 1999) oziroma 0,60 (Gómez in sod., 2009). Primerljiva je bila ocena heritabilitete za višino vihra, merjeno s palico, pri konjih panteneiro pasme v Braziliji, ki je bila ocenje- na na 0,61 (Miserani in sod., 2002). Pri konjih panteneiro pasme v Braziliji (Miserani in sod., 2002) je bila heritabiliteta za širino prsi ocenjena na 0,51, v Španiji pri andaluzijskih konjih (Molina in sod., 1999) na 0,56 oziroma 0,42 (Gómez in sod., 2009) in v našem primeru na 0,43. Na andaluzijskih konjih v Španiji (Molina in sod., 1999; Gómez in sod., 2009) so herita- biliteto za dolžino trupa ocenili na 0,72 oziroma 0,49 (v našem poskusu 0,36) in za obseg piščali na 0,35 oziroma 0,51. Ocene raziskav v Španiji so primerljive z ocenami za populacijo haflinških konj pri nas. Višje ocenjene heri- tabilitete za lastnosti dolžine trupa (0,72) in obseg piščali (0,53) so imeli tudi andaluzijski konji v Španiji (Miserani in sod., 2002). Vse genetske korelacije so bile v našem primeru po- zitivne in ocenjene med 0,38 in 0,95. Genetske korelacije lahko ocenimo kot močne. Visoke genetske korelacije so bile predvsem med višinama vihra in višino križa. Po- dobne rezultate so dobili Suontama in sod. (2009) pri finskih kasačih, kjer je bila genetska korelacija med viši- no vihra in višino križa ocenjena na 0,98 ter med višino vihra in dolžino trupa na 0,84. Molina in sod. (1999) so genetske korelacije ocenili med 0,20 in 0,60. 4 SKLEPI V raziskavi smo proučevali lastnosti zunanjosti pri konjih haflinške pasme v Sloveniji in analizirali podatke, ki so bili zajeti v obdobju zadnjih 19 let. Povprečna sta- rost živali je bila 3,82 leta. Vse ocenjene in/ali izmerjene živali so imele znanega vsaj enega starša in poreklo je bili sestavljeno iz 8 generacij. Najnižja podeljena ocena za la- stnosti zunanjosti je bila 4 in maksimalna 10. Povprečje za skupno oceno je bilo 71,7 točk. Nobena žival se ni uvr- ščala v najboljši 1a kakovostni razred. Razvili smo enostaven model z enim sistematskim in enim naključnim vplivom. Podatki niso omogoča- li proučevanja drugih vplivov, kot so spol in starost. V statistični model bi bilo potrebno vključiti več sistemat- skih vplivov, ki pa morajo biti znani pri večini živali. Za vključitev vplivov bi bilo potrebno izboljšati kakovost podatkov. Naši rezultati so primerljivi z nekaterimi raziskava- mi, ki so jih opravili na haflinških konjih v Italiji in na nizozemskih toplokrvnih konjih. Najnižja heritabiliteta (0,20) je bila ocenjena za globino prsi in najvišja (0,78) za oceno pasemske značilnosti. Za lažjo primerjavo in na- tančnejše ocene parametrov disperzije bi bilo potrebno v analize vključiti več podatkov, uporabljati bi bilo po- trebno celotno ocenjevalno skalo ter ocenjevati oziroma meriti bi bilo potrebno večji delež živali v populaciji, tudi nezaželene. Vredno bi bilo razmisliti tudi o linearni oce- njevalni skali. 5 VIRI Albertsdóttir E., Eriksson S., Näsholm A., Strandberg E., Ár- nason T. 2007. Genetic correlations between competition traits and traits scored at breeding field-test in Icelandic horses. Livest. Sci., 114, 2: 181–187 Arnason T. 1996. Selection criterion for increased long-term response in Nordic-trotters. V: Book of Abstracts of the 47th Annual Meeting of the European Association for Ani- mal Production, Lillehammer. Wageningen, Wageningen Academic Publishers Arnason T., Van Vleck L.D. 2000. Genetic improvement of the horse. Chapter 17th. V: The genetics of the horse. Bowl- ing A.T., Ruvinsky A. (eds.). London, CABI Publishnig: 473–497 Dolvik N., Klemetsdal G. 1999. Conformational traits of Nor- wegian cold-blooded trotters: Heritability and the relation- ship with perfomance. Acta Agric. Scand. Sect. A Anim. Sci., 49: 156–162 Druml T., Baumung R., Sölkner J. 2008. Morphological analy- sis ans effect of selection for conformation in the Noriker draught horse population. Livest. Sci., 115: 118–128 Gómez M.D., Valera M., Molina A., Gutiérrez J.P., Goyache F. 2009. Assessment of inbreeding depression for body mea- surments in Spanish Purebred (Andalusian) horses. Livest. Sci., 122: 149–155 Groeneveld E., Kovač M., Wang T. 1990. PEST, a general pur- pose BLUP package for multivariate prediction and estima- tion. V: 4th World Congress on Genetics Applied to Live- stock Production, Edinburgh, 23–27 July 1990. Edinburgh, The East of Scotland College of Agriculture, 13: 488–491 Acta agriculturae Slovenica, 96/2 – 2010 125 OCENA PARAMETROV DISPERZIJE ZA LASTNOSTI ZUNANJOSTI PRI KONJIH HAFLINŠKE PASME Hayes J.F., Hill W.G. 1981. Modification of estimates of parame- ters in the construction of genetic selection indeces (’Bend- ing’). Biometrics, 37: 483–493 Huganson K. 1994. Breeding of Icelandic toelter horse. Livest. Prod. Sci., 40, 1: 21–29 Jorjani H., Klei L., Emanuelson U. 2003. A simple method for weightet bending of genetic (co)variance matrices. J. Dairy Sci., 86, 2: 677–679 Koenen E., van Veldhuizen A.E., Brascamp E.W. 1995. Genetic parameters of linear scored conformation traits and their relation to dressage and show-jumping performance in the Dutch warmblood riding horses population. Livest. Prod. Sci., 43, 1: 85–94 Kovač M., Groeneveld E., García-Cortés L.A. 2002. VCE-5, a package for the estimation of dispersion parameters. V: Proceedings of the 7th World Congress on Genetics Ap- plied to Livestock Production, Montpellier, 19–23 Aug. 2002. Castanet-Tolosan, INRA, 33: 741–742 Langlois B., Blouin C. 2004. Pratical efficiency of breeding value estimations based on annual earnings of horses for jump- ing, trotting ang galloping races in France. Livest. Prod. Sci., 87, 2–3: 99–107 Langlois B., Vrijenhoek T. 2004. Qualification status and esti- mation of breeding value in French trotters. Livest. Prod. Sci., 89, 2–3: 187–194 Miserani M.G., McManus C., Santos S.A., Silva J.A., Mariante A.S., Abreu U.G.P., Mazza M.C., Sereno J.R.B. 2002. Heri- tability estimates for biometric measures of the Pantaneiro horse. Arch. Zootec., 51: 107–112 Molina A., Valera M., Dos Santos R., Rodero A. 1999. Genet- ic parameters of morphofunctional traits in Andalusian horse. Livest. Prod. Sci., 60, 2–3: 295–303 Pravilnik o ohranjanju biotske raznovrstnosti v živinoreji. Ur.l. RS št. 90-4111/2004: 11001–11009 Rus J. 1996. Preizkus delovne sposobnosti toplokrvnih kobil na tesni postaji Grad Prestranek. Revija o konjih, 7: 7–9 Rus J. 2005. Rejski program za pasmo haflinški konj. Veterinar- ska fakulteta, Ljubljana Samoré A., Pagnacco G., Miglior F. 1997. Genetic parameters and breeding values for linear type traits in the Haflinger horse. Livest. Prod. Sci., 52: 105–111 SAS Inst. Inc. 2001. The SAS System for Windows, Release 8.02. Cary, NC Sobczyinska M., Kownacky M. 1996. Genetic trends in the breeding of Arabian horses. V: Book of Abstracts of the 47th Annual Meeting of the European Association for Ani- mal Production, Lillehammer. Wageningen, Wageningen Academic Publishers Suontama M., Saastamoinen M.T., Ojala M. 2009. Estimates of non-genetic effect and genetic parameters for body mea- sures and subjectively scored traits in Finnhorse trotters. Livest. Sci., 124: 205–209 van Bergen Henk M., van Arendonk J. 1993. Genetic param- eters for linear type traits in Shetland ponies. Livest. Prod. Sci., 36: 273–284 Viliani S. 2008. The Haflinger horse in its country of origin. Haflinger Italija. http://www.haflinger.it/frames/3_haflinger_e.html (1. sep. 2008). Acta argiculturae Slovenica, 96/2, 127, Ljubljana 2010 SUBJECT INDEX BY AGROVOC DESCRIPTORS PREDMETNO KAZALO PO DESKRIPTORJIH AGROVOC Tomaž BARTOL 1 adriatic sea 95–101 highlands 81–86 animal breeding 69–73 himalayan region 81–86 animal morphology 75–80, 111–115, 117–125 horses 117–125 animal population 75–80, 111–115 human nutrition 95–101 antibiotics 81–86 inhibition 87–93 aquatic environment 87–93 isomerization 87–93 autumn 95–101 karst soils 103–109 bacteria 81–86, 87–93 mercury 87–93 behaviour 103–109 microbiology 81–86, 87–93 biogeography 75–80 models 69–73, 81–86 body conformation 75–80, 111–115, 117–125 monitoring 87–93 body measurements 75–80, 111–115, 117–125 nigeria 75–80 body parts 75–80, 111–115, 117–125 nutritive value 95–101 body regions 75–80 oncorhynchus mykiss 111–115 breeds (animals) 75–80, 117–125 pastures 103–109 cattle 75–80 phenotypes 75–80 chromosome number 111–115 proximate composition 95–101 common salt 103–109 pseudomonas putida 87–93 consumption 103–109 recommended dietary allowances 95–101 data analysis 69–73 resistance to chemicals 81–86 diet 95–101 rotational grazing 103–109 dietary guidelines 95–101 sampling 69–73 dimensions 75–80 sardina 95–101 diploidy 111–115 sardines 95–101 drinking habits 103–109 seasons 95–101 ecosystems 87–93 selection 69–73, 75–80, 117–125 environmental factors 81–86 sheep 103–109 environmental protection 87–93 simulation models 69–73 fatty acids 95–101 soil 81–86 feeding habits 103–109 statistical data 69–73, 75–80 fish 95–101, 111–115 statistical methods 69–73, 75–80 foods 95–101 survival 111–115 genetic correlation 117–125 therapeutic diets 95–101 genetic distance 75–80, 111–115 triploidy 111–115 genetic parameters 75–80, 111–115, 117–125, 103–109 viability 111–115 grazing systems 103–109 water pollution 87–93 growth 87–93 winter 95–101 heritability 117–125 1 Univ. of Ljubljana, Biotechnical Fac, Dept. of Agronomy, Jamnikarjeva 101, SI-1000 Ljubljana, Slovenia, Assoc.Prof., Ph.D., e-mail: tomaz.bartol@bf.uni-lj.si Acta argiculturae Slovenica, 96/2, 129, Ljubljana 2010 SUBJECT INDEX BY AGRIS CATEGORY CODES VSEBINSKO KAZALO PO PREDMETNIH KATEGORIJAH AGRIS Nataša SIARD 1 1 Univ. of Ljubljana, Biotechnical Fac., Dept. of Animal Science, Groblje 3, SI-1230 Domžale, Slovenia, Ph.D., M.Sc., e-mail: natasa.siard@bf.uni-lj.si Animal husbandry – L01 75–80, 103–109 Animal genetics and breeding – L10 117–125 Aquatic ecology – M40 111–115 Water resources and management – P10 87–93 Soil science and management – P30 81–86 Soil biology – P34 81–86 Food composition – Q04 95–101 Pollution – T01 87–93 Mathematical and statistical methods – U10 69–73 Acta argiculturae Slovenica, 96/2, 131–132, Ljubljana 2010 ABECEDNO KAZALO AVTORJEV AUTHOR’S INDEX Št. No. Avtor Author Stran primarnega prispevka Page of the primary source 1. AMUSAN Samuel 75–80 2. BARTOL Tomaž 127 3. BISTAN Mirjana 87–93 4. BOJKOVSKI Daniela 103–109 5. DOVČ Peter 67–68 6. FLISAR Tina 69–73 7. GARCÍA-CORTÉS Luis Alberto 69–73 8. GAŠPERLIN Lea 95–101 9. GORJANC Gregor 69–73 10. HARUNA Hadiza Salihu 75–80 11. IDAHOR Kingsley Omogiade 75–80 12. KOMPAN Dragomir 103–109 13. KOVAČ Milena 117–125 14. MALOVRH Špela 117–125 15. MARIN Monika 95–101 16. MARINŠEK LOGAR Romana 87–93 17. MARTÍNEZ-ÁVILA Jose Carlos 69–73 18. PLANINC Martina 117–125 19. POHAR Jurij 111–115 20. POLAK Tomaž 95–101 21. RUS Janez 117–125 22. SIARD Nataša 129 23. STRES Blaž 81–86 24. STRGAR Klavdija 111–115 25. ŠTUHEC Ivan 103–109 Acta agriculturae Slovenica, 96/2 – 2010132 Št. No. Avtor Author Stran primarnega prispevka Page of the primary source 26. VODOVNIK Maša 87–93 27. WHETO Matthew 75–80 28. YAKUBU Abdulmojeed 75–80 29. ZOREC Maša 87–93 30. ZUPAN Manja 103–109 31. ŽLENDER Božidar 95–101 Acta argiculturae Slovenica, 96/2, 133–134, Ljubljana 2010 NAVODILA AVTORJEM PRISPEVKI Sprejemamo izvirne znanstvene članke, predhodne objave in raziskovalne notice s področja zootehnike (ge- netika, mikrobiologija, imunologija, prehrana, fiziologi- ja, ekologija, etologija, mlekarstvo, ekonomika, živalska proizvodnja in predelava živalskih proizvodov, tehno- logija in dokumentalistika) v slovenskem in angleškem jeziku, pregledne znanstvene članke pa samo po poprej- šnjem dogovoru. Objavljamo tudi prispevke, podane na simpozijih, ki niso bili v celoti objavljeni v zborniku sim- pozija. Če je prispevek del diplomskega, magistrskega ali doktorskega dela, navedemo to in tudi mentorja v sprotni opombi na dnu prve strani. Navedbe morajo biti v slo- venskem in angleškem jeziku. Pri prispevkih v slovenskem jeziku morajo biti pre- glednice, grafikoni, slike in priloge dvojezični, povsod je slovenščina na prvem mestu. Naslovi grafikonov in slik so pod njimi. Preglednice, slike in grafikoni so v besedilu. Grafikoni morajo biti črno-beli. Latinske izraze pišemo ležeče. V slovenščini uporabljamo decimalno vejico, v angleščini decimalno piko. Prispevki naj bodo strnjeni, kratki, največ 12 strani, napisani z urejevalnikom besedil in oddani v doc ali rtf formatu (Windows). Izgled strani naj bo čim bolj eno- staven; v besedilo ne vstavljajte glave in noge. Pisava v besedilu in preglednicah je Times New Roman, velikost črk 12, v obsežnih preglednicah je lahko 10, pisava v gra- fikonih in slikah je Ariel, velikost črk najmanj 8, pisava za primerjave nukleotidnih in aminokislinskih zaporedij je Courier; zunanji rob 2,0 cm, notranji 2,5 cm. PRVA STRAN Na prvi strani prispevka na desni strani označimo vrsto prispevka, sledi naslov prispevka, pod njim avtorji. Ime avtorjev navedemo v polni obliki (ime in priimek). Vsakemu avtorju dodamo sprotno opombo, ki je vidna na dnu strani, in vsebuje polni naslov ustanove ter znan- stveni in akademski naslov; vse v jeziku prispevka. Na- vedemo sedež ustanove, kjer avtor dela. Če je raziskava opravljena drugje, avtor navede tudi sedež te inštitucije. Na željo avtorjev bomo navedli naslov elektronske pošte. Pod imeni avtorjev je datum prispetja in datum sprejetja prispevka, ki ostaneta odprta. Sledi razumljiv in poveden izvleček z do 250 besedami. Vsebuje namen in metode dela, rezultate, razpravo in sklepe. Sledijo ključne besede. Izvlečku v jeziku objave sledi naslov in izvleček s ključnimi besedami v drugem jeziku. VIRI V besedilu navajamo v oklepaju avtorja in leto ob- jave: (priimek, leto). Če sta avtorja dva, pišemo: (priimek in priimek, leto), če je avtorjev več, pišemo: (priimek in sod., leto). Sekundarni vir označimo z »navedeno v« ali »cv.«. Seznam virov je na koncu prispevka, neoštevilčen in v abecednem redu. Vire istega avtorja, objavljene v istem letu, razvrstimo kronološko z a, b, c. Primer: 1997a. Ne- kaj primerov navajanja virov: Vodovnik M., Marinšek-Logar R. 2008. Način delovanja in učinki probiotikov v prehrani živali. Acta agriculturae Slo- venica, 92, 1: 5–17 Acta agriculturae Slovenica, 96/2 – 2010134 Fraser A.F., Broom D.M. 1990. Farm animal behaviour and wel- fare. London, Bailliere Tindall: 437 str. Hvelplund T. 1989. Protein evaluation of treated straws. V: Evaluation of straws in ruminant feeding. Chenost M., Rei- niger, A. (ur.). London, Elsevier Applied Science: 66–74 Žgajnar J., Kermauner A., Kavčič S. 2007. Model za ocenjevanje prehranskih potreb prežvekovalcev in optimiranje krmnih obrokov. V: Slovensko kmetijstvo in podeželje v Evropi, ki se širi in spreminja. 4. konferenca DAES, Ljubljana, 8–9 sep. 2007. Kavčič S. (ur.). Domžale, Društvo agrarnih eko- nomistov Slovenije: 279–288 ISO 5534 / IDF 4. Cheese and processed cheese – Determina- tion of the total solids content – Reference method. 2004: 1–7 Frajman P., Dovč P. 2004. Milk production in the post-genomic era. Acta agriculturae Slovenica, 84, 2: 109–119. http://aas.bf.uni-lj.si/zootehnika/84-2004/PDF/84-2004-2- 109-119.pdf (15. mar. 2009) ODDAJA Avtorji prispevke oddajo v natisnjenem in elektron- skem izvodu. Priložijo tudi izjavo s podpisi vseh avtorjev, da avtorske pravice v celoti odstopajo reviji. Prispevke recenziramo in lektoriramo. Praviloma pošljemo mnenje prvemu avtorju, po želji lahko tudi drugače. Če urednik ali recenzenti predlagajo spremem- be oz. izboljšave, vrne avtor popravljeno besedilo v 10 dneh v natisnjenem in elektronskem izvodu. Ko prvi av- tor vnese še lektorjeve pripombe, odda popravljeno bese- dilo v natisnjenem in elektronskem izvodu. Pri oddaji končne verzije avtor priloži jasno označe- ne izvirnike slik (ločene grafične datoteke ali fotografije). Datoteke slik poimenuje enako kot v tekstu (npr. Slika1. jpg, Slika2.eps, Slika3.bmp). Originalne fotografije na av- torjevo željo vrnemo. Vektorske slike sprejemamo samo v eps (Encapsulated Postscript) formatu, s tekstom, ki je spremenjen v krivulje. Rasterske slike morajo biti v enem od običajnih formatov (npr. tiff, jpg, bmp). Ločljivost naj bo vsaj 300 dpi. Prispevke sprejemamo vse leto. Acta argiculturae Slovenica, 96/2, 135–136, Ljubljana 2010 NOTES FOR AUTHORS PAPERS We publish original scientific papers, preliminary communications and research statements on the subject of animal science (genetics, microbiology, immunology, nutrition, physiology, ecology, ethology, dairy science, economics, animal production and food processing, technology and information science) in Slovenian and English languages while scientific reviews are published only upon invitation. Reports presented on conferences that were not published entirely in the conference reports can be published. If the paper is part of BSc, MSc or PhD thesis, this should be indicated together with the name of the mentor at the bottom of the front page and will appear as foot note. All notes should be written in Slov- enian and English language. Papers in Slovenian language should have tables, graphs, figures and appendices in both languages, Slove- nian language being the first. Titles of graphs and figures are below them. Figures and graphs are part of the text. Clearly marked original figures should be added (pho- tographs or separate graphic files); they can be returned upon request. Latin expressions are written in italics. Decimal coma is used in Slovenian and decimal point in English. The papers should be condensed, short and should not exceed 12 pages, editted with word processor and submitted as doc or rtf file (Windows). Text formatting should be as simple as possible, without headers and footers. Font Times New Roman, size 12 should be used for text and tables (in large tables size 10 is allowed), Ariel should be used for graphs and figures (letter size at least 8) and Courier for nucleic- and amino acid sequence alignments. Right margin is 2.0 cm, left margin 2.5 cm FIRST PAGE The type of the paper should be indicated on the first page on the right side following by the title of the pa- per and authors. Full names of the authors are used (first name and surname). Each name of the author should have been added an index, which is put immediately after the author’s name and displayed in the footnote. It contains address of the institution and academic degree of the author, in the language of the paper. The address of the institution in which the author works is indicated. If the research was realised elsewhere, the author should name the headquarters of the institution. E-mail is op- tional. Under the address of the authors some space for dates of arrival and acceptance for publishing should be left. A comprehensive and explicit abstract up to 250 words follows indicating the objective and methods of work, results, discussion and conclusions. Key words fol- low the abstract. The abstract in the language of the paper is followed by the title, abstract and key words in the alternative lan- guage. REFERENCES References should be indicated in the text by giv- ing author’s name, with the year of publication in paren- theses, e.g. (surname, year). If there are two authors, the following form is used: (surname and surname, year). If there are more than two authors, we use (surname et al., year). Secondary sources should be quoted in the form “cited in”. The references should be listed at the end of Acta agriculturae Slovenica, 96/2 – 2010136 the paper in the alphabetical order and not numbered. If several papers by the same author and from the year are cited, a, b, c, etc. should be put after the year of the publication: e.g. 1997a. Some examples: Simončič M., Horvat S., Stevenson P.L., Bünger L., Holmes M.C., Kenyon C.J., Speakman J.R., Morton N.M. 2008. Di- vergent physical activity and novel alternative responses to high fat feeding in polygenic fat and lean mice. Behavior Genetics, 38, 3: 292–300 Fraser A.F., Broom D.M. 1990. Farm animal behaviour and wel- fare. London, Bailliere Tindall: 437 p. Hvelplund T. 1989. Protein evaluation of treated straws. In: Evaluation of straws in ruminant feeding. Chenost M., Rei- niger, A. (ed.). London, Elsevier Applied Science: 66–74 Žgajnar J., Kermauner A., Kavčič S. 2007. Model za ocenjevanje prehranskih potreb prežvekovalcev in optimiranje krmnih obrokov. In: Slovensko kmetijstvo in podeželje v Evropi, ki se širi in spreminja. 4. konferenca DAES, Ljubljana, 8–9 sep. 2007. Kavčič S. (ed.). Domžale, Društvo agrarnih eko- nomistov Slovenije: 279–288 ISO 5534 / IDF 4. Cheese and processed cheese – Determina- tion of the total solids content – Reference method. 2004: 1–7 Frajman P., Dovč P. 2004. Milk production in the post-genomic era. Acta agriculturae Slovenica, 84, 2: 109–119. http://aas.bf.uni-lj.si/zootehnika/84-2004/PDF/84-2004-2- 109-119.pdf (15. mar. 2009) DELIVERY Papers should be delivered as a printed and elec- tronic copy. A statement signed by all authors transfers copy rights on the published article to the Journal. Papers are reviewed and edited. First author re- ceives a review if not defined otherwise. If reviewers sug- gest some corrections, the author should forward them within 10 days in printed and electronic form. After the first author considers the referee’s notes, the corrected paper should be sent in printed and electronic form to the Editor. Submission of the final version must contain prop- erly labelled original figures (separate files or photogra- phies). The figure files should be labelled as they appear in the text (Figure1.jpg, Figure2.eps, Figure3.bmp). Orig- inal photographies can be returned to the author upon request. Vector graphics have to be in eps (Encapsulated Postscript) format with the text transformed in curves. Raster figures and photos should be in one of common formates (e.g. tiff, jpg, bmp) with at least 300 dpi resolu- tion. Papers are accepted all the year.