Restricted Maximum Likelihood to estimate variance components for mixed models with two random factors

Protection of Concrete Structures: Performance Analysis of Different Commercial Products and Systems

The increasing demand for reconstructions of concrete structures and the wide availability on the market of surface protective products and systems could lead to misunderstandings in the decision of the most effective solution. Surface protectors have become increasingly widespread in recent years in concrete restoration interventions thanks to their properties: they are able to protect the substrate from aggressive agents and consequently extend the useful life of the structures. The aim of this article is first of all to present the surface protective treatments available on the market, outlining their strengths and weaknesses. Subsequently, a characterization of seven different commercial coatings for reinforced-concrete structures is provided, taking into account chemical nature, fields of use and effectiveness, both in terms of physic and elastic performance and resistance to aggressive agents that undermine the durability of the treated concrete elements.

Comparison of two models for estimation of variance components in a sample of Spanish Holstein Friesians

SUMMARY

Two models for estimating genetic parameters in animal populations were compared. The cheapest, in terms of computing costs, was based on a sire model and the most expensive on an animal model. In order to check the accuracy of both methods they were applied to the best available sample of Spanish Friesian cattle. Milk production of each cow, as well as fat contents were simulated using prior estimates of genetic additive, permanent environment and residual variances, keeping the actual mating scheme, so that the simulation mimicked the actual population structure. Results, under the above premises, show no advantages of the more costly procedure over the cheaper one, suggesting that a sire model based estimating procedure for genetic parameters may be preferred when a small number of individuals, little pedigree information and highly disequilibrated distribution of effects characterize the data. ZUSAMMENFASSUNG: Vergleich zweier Methoden der Varianzkomponentenschätzung beim Spanischen Holstein Zwei Modelle zur Schätzung von genetischen Parametern wurden verglichen. Das im Hinblick auf die Rechenleistung günstigere Modell war ein Vatermodell, das aufwendigere Modell ein Tiermodell. Um die Genauigkeit der Modelle in der Praxis zu prüfen, wurden sie auf verfügbares Material der spanischen Holstein-Friesian Population angewendet. Sowohl die Milchleistung jeder Kuh als auch der Fettgehalt wurden simuliert mit Hilfe von Schät- zwerten der additiv-genetischen Varianz, der permanenten und der residuellen Umweltvarianz. Dabei wurde das tatsächliche Ungleichgewicht in der Population berücksichtigt. Unter den obengenannten Voraussetzungen zeigen die Ergebnisse keinen Vorteil der rechnerisch aufwendigeren Prozedur gegenäber der einfacheren. Die Schlußfolgerung lautet, daß das Vatermodell unter den Bedingungen der kleinen Tierzahlen, wenig Pedigree-Informationen und einer stark unbalan- cierten Verteilung der Tiere auf die einzelnen Effektklassen zu bevorzugen ist.

MIVQUE estimation of genetic parameters for traits of dam and young traits in a prolific species

SUMMARY

A two-trait model is proposed to analyse jointly a dam trait and a young trait in prolific species such as pigs or rabbits. The model is a true repeatability animal model for the dam trait and a sire-dam model with maternal effects for the young trait. A feasible MIVQUE is worked out to estimate the variance-covariance components involved in the model. In order to reduce demands on computer memory and time an equivalent model is proposed and some algebraic work has been done. Thus, the time required to compute the expectations of the MIVQUE quadratics is only 70% of the one necessary to invert the coefficient matrix. Thus REML could be solved through an I-MIVQUE algorithm as an alternative to the EM algorithm. The suggested procedure presumes that the levels of the fixed effects are low. This is frequently the case when selecting pigs or rabbits in a nucleus. Finally, an example is presented for the analysis of litter size at weaning and weight at 77 days in a line of rabbits. ZUSAMMENFASSUNG: MIVQUE-Schätzung genetischer Parameter für Merkmale von Müttern und Jungen fruchtbarer Haustierarten Es wird ein Zwei-Merkmal-Modell verwendet, das ein echtes Wiederholbarkeitsmodell für Muttermerkmal und ein Vater/Muttermodell mit Maternalwirkungen für Nachkommenmerkmale darstellt. MIVQUE wird zur Schätzung der Varianzkomponenten verwendet. Zur Verminderung von Ansprüchen an Speicher und Zeit wird ein äquivalentes Modell vorgeschlagen und algebraische Untersuchungen durchgeführt. Die für die Berechnung der Erwartungswerte von MIVQUE notwendige Zeit beträgt 70%, der für Inversion der Koeffizientenmatrix notwendig ist. REML konnte mit I-MIVQUE-Algorithmus als Alternative zu EM-Algorithmus gelöst werden. Die vorgeschlagene Methode verlangt ein niedriges Ausmaß von fixen Wirkungen, was häufig für Nukleus-selektierte Schweine oder Kaninchen zutrifft. Ein Beispiel analysiert Wurfgröße beim Absetzen und 77-Tage-Gewicht von Kaninchen.

Estimates of repeatability of reproductive measures in Canadian Holsteins

Repeatabilities and heritabilities of days to first service, days open, and number of services per conception were estimated from 235,589 records on 80,333 Holstein cows, daughters of 306 sires obtained from the Quebec Dairy Herd Analysis Service, by REML. The model for statistical analyses included herd-year-season as a fixed effect and sire, cow (sire's daughter) within sire, and error as random effects. Variance components corresponding to sire, cow within sire, and error were then estimated by REML. Heritability estimates, obtained from the sire variance component, were .03, .05, and .03 for days to first service, days open, and number of services per conception, respectively. Repeatability estimates, obtained from the sire and cow within sire variance components were .08, .10, and .07 for days to first service, days open, and number of services per conception, respectively.

The importance of maternal lineage on milk yield traits of dairy cattle

Maternal lineage effects on milk yield traits, considered indicative of cytoplasmic inheritance, were evaluated with animal models. Cattle were from a selection experiment begun in 1968. Maternal pedigrees were traced to the first female member in the Holstein-Friesian Herdbook; purchased cows entering the herd, considered foundation females, were assigned to maternal lineage groups. All models accounted for year-season of calving, parity, and selection lines. Maternal lineage effects were included in a repeated records model with cow effects and preadjustment for sire and maternal grandsire transmitting abilities. Maternal lineage accounted for 5.2, 4.1, and 10.5% of phenotypic variation of preadjusted records of milk yield, fat yield, and fat percentage, respectively. Maternal lineage was evaluated as a fixed effect in an animal model including random animal and permanent environmental effects. Maternal lineage significantly affected fat percentage but not milk yield. Maternal genetic (nuclear) effects and their covariance with additive animal effects did not significantly account for additional variation nor did they influence maternal lineage estimates. Maternal lineage affected calculated net energy of milk but was not important for SNF yield or concentration. Maternal lineage influenced fat percentage, energy concentration, and, to a lesser extent, fat yield in milk of dairy cattle.

Genetic parameters for yield traits of Holsteins, including lactose and somatic cell score

First lactation milk, fat, protein, and lactose yields and percentage yields were analyzed using a multiple-trait sire model including herd-year-season, sire group, and age of cow as fixed effects. Somatic cell score was fit both as a fixed effect in the model and as an additional dependent variable in two analyses with almost identical results. Variance components were estimated using REML with an expectation-maximization algorithm including sire relationships. Lactose means ranged from 4.84 to 4.97% across three dairy breeds. Data used to estimate variance components were first lactation Holstein records collected from 1986 to 1988 from 5246 daughters of 392 AI sires. Heritability estimates were .30, .29, .27, and .26 for milk, fat, protein, and lactose yields; .45, .47, and .53 for fat, protein, and lactose percentage yields; and .16 for somatic cell score. Genetic correlations of lactose percentage with milk, fat, protein, fat and protein percentages, and somatic cell score were -.30, -.16, -.21, .10, .29, and -.11, respectively, and phenotypic correlations were -.08, -.02, .01, .11, .29, and -.15.

Estimation of repeatability of calving ease in Canadian Holsteins

Restricted maximum likelihood was used to estimate repeatabilities and heritabilities for calving ease from a data set of 636,972 Holstein parturitions obtained from the Quebec Dairy Herd Analysis Service. Subsets of the data, chosen to reflect different stages of cow maturity, were analyzed separately and compared. Initial least squares analyses were used to define the fixed effects to include in the model. The effects of herd-year-season, sex of calf, parity, and sex by parity interaction were significant in all subsets. Age and parity by age interaction were significant in subsets with primarily younger animals. Variances for sire of calf, cow, and residual effects were then estimated by REML. Repeatabilities ranged from 5.98 to 7.87% and heritabilities for calving ease as a direct effect ranged from 1.97 to 4.70%.