Dairy cattle genetics and its applications in Brazil

F E Madalena*, M G C D Peixoto** and J Gibson***

* Independent consultant,
www.fernandomadalena.com
contact@fernandomadalena.com
**EMBRAPA-Gado de Leite, R. Eugênio do Nascimento 610
36038-330 Juiz de Fora-MG, Brazil
***Centre for Genetic Analysis and Applications, School of Environmental and Rural Science,
University of New England, Armidale NSW 2351, Australia

Abstract

The Brazilian dairy industry is based mainly on pasture-oriented, dual purpose systems in the tropical part of the Country, utilizing B. taurus x B. indicus hybrid animals, mostly Holstein/Gir, milked with restricted suckling of calves, which was shown to be more economical than artificial rearing in most circumstances. The Brazilian farmers have maintained for decades the dairy herd intermediate between B. taurus and B. indicus. Results of surveys and census data reveal a wide variation in dairy farm size, which is associated with different genotype use; low-input, smaller farms use a higher proportion of the more resilient, low B. taurus grade cows, while larger farms use more of the higher yielding, higher B. taurus grades.

A trial comparing contemporaneous females (527) of six Holstein-Friesian (H) x Guzerá crosses  (¼ to ≥31/32/ H) milked twice a day with the calf stimulus in 67 farms showed important favourable heterosis in milk solids yield, fertility, mortality, herd life, weight, weight/height ratio and tick resistance, leading to an overwhelming importance of heterosis for profitability, particularly in lower input farms. Profit per day of herd life was higher for F₁ than for other crossbreeding strategies. In low management farms crisscrossing attained predicted 59% of the F₁ profit, ⅝ inter se 30% and ≥31/32 H (upgrading to H) -21%, i.e. were uneconomic. In better managed farms a strategy of repeating the H sires for two generations followed by one generation by Zebu sires attained 75% of the F₁ profit, the same as the ≥31/32 H, while the ⅝ H inter se showed a negative profit of -18%. Hence, new breeds developed from crossbred foundations would suffer from loss of  heterosis.

Subsequent analyses of data from institutional or private herds supported those conclusions except for cows mechanically milked and artificial calf rearing, in which milk yield increased with higher Holstein gene fractions, although economic analyses have not been conducted in such systems. The published evidence does not support the widely held breeders’ belief that a ⅝ inter se Holstein composite is an optimal goal. For systems in which the male calves were not reared, three way Jersey/H/Gir crosses were more profitable than H/Gir and Brown Swiss/H/Gir on account of their higher fertility, lower weight and longer herd life.

Progeny testing programmes in the Gir, Guzerá and Girolando breeds are described along with an associated embryo transfer full-sib selection programme in the Guzerá. These programmes have given great commercial impact to those breeds. Holstein genes flow into Brazil from abroad, mostly via imported semen which accounts for 92% of the total semen sold of this breed. Very little genetic progress has been made for milk yield in this breed despite the high genetic correlation between yield in Brazil and in semen exporting countries and fertility and health problems have been reported. 

Extensive research is conducted in Minas Gerais State on production systems based on F₁ B. taurus x B. indicus hybrids. Milk yield per day of calving interval in H x Gir, H x Guzerá and H x Nelore F₁ crosses was 7.1, 6.4 and 5.6 kg/d, respectively, and weight at first calving 448, 466 and 461 kg.

A strategy of continuous replacement with F₁ B. taurus x B. indicus heifers may be very profitable depending on price and availability of such animals and several current commercial developments in the production of F₁s, either by artificial insemination or by ovum pick up and in vitro fertilization with sexed semen are reviewed.

Key words: Tropical dairy production, crossbreeding strategies, heterosis, Gir, Guzerá, Girolando

Introduction

This review is prompted by the strong interest in tropical dairy development and use of appropriate genotypes in many developing countries. Brazil has extensive experience in tropical dairy systems with the unique perspective of having systems with very low through to high feed quality and supply, coupled with institutions and national will to undertake research and development of tropically adapted dairy cattle genotypes and production systems. Brazil also has an advanced commercial breeding sector and expertise in traditional and advanced reproductive technologies, that is available to other tropical countries interested in developing their dairy industries. The current review presents a critical analysis of what has been learned in Brazil about tropically adapted dairy genotypes that should be relevant to other countries. The production systems context of dairy genetics in Brazil is explained and key results of production systems research that are relevant to application in tropical developing countries are also highlighted.

Overview of Brazilian dairy development

An overview of the Brazilian dairy industry is presented here and further background is given in the Appendix. Dairy production is an important agricultural activity in Brazil, showing continuous growth and evolution in the recent decades. It occurs in practically all the national territory (554 of the country’s 558 micro-regions, Zoccal and Carneiro 2008). Brazil ranks as the fifth largest cow milk producer country, after the USA, India, China and Russia. The gross value of milk production in 2009 was US$ 10,019 million, ranking fifth among national agricultural products (FAOSTAT 2011). Between 2000 and 2008, the annual growth rate of cow milk production was 4.0% while the global rate was 2.1% (Carvalho 2010). Milk production tripled between 1975 and 2010, from 10.0 to 31.7 million tonnes/y, with 12.3 and 22.9 million milked cows, yielding 811 and 1381 kg milk/cow/y (FAOSTAT 2011). Between 1975 and 2010 the population increased from 108.4 to 194.9 million people (World Bank 2011). Annual milk consumption increased from 69 to 161 kg/capita (MilkPoint 2011), mainly due to higher income, particularly in the low income strata where income-elasticity is highest (e.g. Pinha et al 2010). Control of inflation and public policies on re-distribution of income contributed to the enhanced consumption. The Brazilian dairy industry is basically geared to the domestic market, although it has great exporting potential. Previously a net importer, Brazil achieved close to zero import-export balance in 2004, with fluctuations over time due to the currency exchange rate and other reasons, such as the policy to import from the Mercosul neighbours Argentina and Uruguay. In 2010, the milk equivalent of the net import-export balance was 0.6 million tonnes (MilkPoint 2011), equivalent to 1.8% of the total national milk production. The Brazilian dairy industry success story is based mainly on pasture-oriented, dual purpose systems in the tropical part of the Country, utilizing  B. taurus x B. indicus hybrid animals, one of the world largest crossbred dairy cattle herd.

Production systems

Nutrition and health

Milk production in Brazil is predominantly based on pastures supplemented with concentrate feeds. In the Southeast and Center-West regions roughages are fed in the dry season (generally April to September). Pasture irrigation and rotational grazing techniques are applied by a minority of farms. Some of the main cultivated pastures are Green panic (Panicum maximum Jacq. cv Tanzania and Mombaça), Cynodon spp, c.v. Tifton, elephant grass (Pennisetum purpureum Schum. cv Napier), Brachiaria brizantha c.v. Marandu and Brachiaria decumbens c.v. Basilisk). The main concentrate supplements are maize, sorghum, soybean meal and cottonseed. During the dry season feeding sugar cane-urea and zero-grazed elephant grass are common and to a lesser extent maize and sorghum silage and dehydrated citrus pulp.

Vaccinations against foot and mouth disease and brucellosis are compulsory and vaccinations against clostridiosis and rabies are common. Ticks (Ripicephalus microplus) and gastrointestinal parasites are important but control is very variable and usually relies on chemicals rather than on strategic plans.

Restricted-suckling, dual purpose systems

In tropical Brazil males are generally reared as they are an important source of receipts. This does not apply to the more specialized systems using B. taurus or high grades, in which males, lacking adaptation, are sold as bobby calves. Dual purpose systems based on restricted suckling of calves allow farmers more flexibility to adapt to the wide fluctuations of milk and beef prices. Farmers regulate the amount of milk produced by altering the amount of concentrates fed, by milking only once a day on part or the whole herd leaving more milk for the calves, or by switching to a total suckling system (Madalena 2001a).

B. indicus cows and their low grade B. taurus crosses need to be stimulated by the presence of the calf for milk ejection when mechanically milked. Oxytocin release was considered sufficient for machine milking of Gir and Holstein/Gir crosses in some experiments but not in others and contradictory results were also obtained in relation to the importance of inhibitory cortisol levels (e.g. Porcionato et al 2005, Negrão 2008).

Making use of the calf to stimulate milking is a common practice in tropical Latin-America, including Brazil. For example, Madalena et al (1997a) reported that this system was used by 95% of 300 farms surveyed in the State of Minas Gerais. The usual practice in Latin-America is that the calves are kept in or brought to a pen adjacent to the milking parlour and when a cow is about to be milked her calf is brought to her, allowed to suckle a few milk jets and then tied up in front of her or to her front leg during milking after which both dam and calf remain together in another pen/paddock until milking of the whole herd is completed. Calves and cows are then separated and moved to their own pastures/paddocks/pens until the next milking. Although it is now well established that the restricted suckling system coupled with early weaning results in higher total milk production (i.e. saleable plus consumed by the calf) than milking without the calf (e.g. Preston and Vaccaro 1989, Combellas, 2006), it is often considered a backward practice on the grounds that milking with the calf present is cumbersome and it would cause poor milk hygiene. However, Junqueira et al (2005) showed that in a 120 cow herd the additional time spent milking 2X with restricted suckling (for two months and then weaned) was similar to bucket feeding/cleaning in artificial calf rearing (3.8 and 3.6 min/calf/day). So the extra milk sold economically justified the first system. This result applies to the bucket machine-milking parlour and calf huts in that experiment and may not extrapolate to other facilities. No significant differences were found between treatments in somatic cell count, calf mortality and morbidity, 2-month calf weight, cow weight and body condition score, calving to first oestrus and calving to conception intervals. The mean weighted difference between the two milking systems in calving to first oestrus intervals in nine literature results was 8.5 + 11.6 days (Junqueira et al 2005).  On the other hand, under poor animal health standards restricted suckling reduces mastitis incidence and calf mortality. Benefits due to the calf removal of residual milk have been reported also for B. taurus cows (Caldas and Madalena 2001).

Some large herd managers prefer artificial rearing to speed up milking, which implies a  trade off of less milk production for faster milking speed (which most farmers are probably unaware of) and some farmers are resorting to oxytocin daily injections to mechanically milk hybrid cows without the presence of the calf. The whole field of milking systems, including mechanical facilities to milk in the presence of the calf, is an important aspect of tropical dairy production where further research and extension is still needed

Farm profiles  

The results of a survey based on interviewing 1000 farm managers in Minas Gerais in 2005 (Table 1) illustrate the wide variation in farm size and management. Average land area for dairying was 57 ha, 86% of which was occupied with pastures (natural or cultivated) and the rest with sugar cane and elephant grass stands and to a lesser extent maize or sorghum silage. Seventy one percent of the capital invested corresponded to land. Most (94%) farms had electricity supply and 89% had a road allowing access of refrigerated milk trucks. Milk cooling tank is present in most farms except the smaller ones, which generally use a collective tank for several neighbouring farms. Machine milking was commonest in the largest farms.

Practically all farms, except in the lower production class, fed concentrates. Bovine somatotrophin (BST) was used in 15% of farms.

Family labour prevailed (62% of total) in the small farms selling <50 l/d but it was only 16% in the largest farms selling >1000 l/d. However, almost all farms (97%) were managed by the owner, with low frequency of hired managers in all farm sizes.

Artificial calf rearing was uncommon (2%) in the smaller farms and increased up to 50% in the largest farms. Animals sold were responsible for 19% of total receipts. Males made up 28% of the herd and were sold mostly at weaning or as yearlings.  

Breeds

B. taurus breeds, mainly Holstein, are used in the South Region, or, roughly, South of the Tropic of Capricorn, while mainstream dairy B. taurus x B. indicus crosses prevail North of it, as they combine high milk yield of the former with adaptation of the latter species to coarse pastures/roughages, dry seasons, parasites and diseases and the direct climatic effects on the animals. The direct climate effects by no means are the main restrictive factor to using purebred Holsteins or other specialized B. taurus breeds, and hybrids prevail even in areas where the temperature-humidity index (THI) is below 72  (e.g. in São Paulo and Minas Gerais, Peixoto et al 2011). Azevedo et al (2005) reported that ½, ¾ and ⅞ Holstein milking cows showed critical THI values for normothermia of 80, 77 and 75, respectively

Herdbooks are kept by breeders associations on behalf of the Ministry of Agriculture. In addition, an all B. indicus breed registry is regulated by the Brazilian Association of Zebu Breeders (ABCZ) which congregates the individual breed associations. In 2010 the Gir registered 25,530 animals and the Guzerá 17,570 (L.A. Josahkian, ABCZ, personal comunication). Both breeds have a single herd book, although dairy and dual purpose lines exist in the Gir and dual purpose and beef lines in the Guzerá. The Dairy Gir Breeders Association (ABCGIL) initially formed to promote recording, was instrumental in the establishment of the breed progeny testing programme (see below). The genetic structure of the Brazilian Zebu breeds was described by Faria et al (2009).

The Girolando (Gir-Holland) was founded in 1978 to develop an inter se mated ⅝ Holstein ⅜ Gir composite (paradoxically named ”pure synthetic”), using traditional old fashioned crossing based on the belief that a ⅝ B. taurus fraction is necessary in new breeds, although  this belief has not been ever scientifically confirmed in Brazil nor elsewhere. All crossbred generations (¼, ½, ¾, etc) are registered. The word Girolando may generate confusion, because it is used to refer to any crossbred animals used for dairy production as well as for the registered elite of this breed.

Holstein, Jersey and minority B. taurus breeds genes flow in from abroad via imported semen and other genetic materials. The Holstein Association registered 52,240 animals in 2011 (A. Marques, Brazilian Holstein Breeders Association, personal communication).  There are many other minority breeds, described by Mariante et al (2003).

There are no cattle census data by breed, so information on this topic comes from specific sample surveys. As an example, a survey involving more than 7,000 cows in farms affiliated to the main dairy processor in the State of Minas Gerais showed 89% having B. taurus x B.indicus genetics (Figure 1).

There is an association of the B. taurus grade with the input level of the production system. Although intermediate grades prevail at all input levels, low-input, smaller farms, use a higher proportion of the more resilient, low B. taurus grade cows, while larger farms use more of the higher yielding, higher B. taurus grades (e.g. Figure 1).

As may be seen in Figure 2, the Holstein and Gir breeds are widely preferred, being used respectively on 94 and 80% of the farms surveyed.  In the above mentioned 1,000 farm survey in Minas Gerais, 67% of farms used B.taurus x B.indicus hybrids, 25% used “predominantly B. taurus" and 8% “predominantly B. indicus” (FAEMG 2006). A similar 2009 survey on 500 dairy farms in the State of Goiás indicated 75% of the cows between ¼ and ¾ and 22%  >¾ B. taurus (FAEG 2009). The trend of somewhat higher B. taurus grades in the larger farms was also present in the latter two surveys.

Crossbreeding practices

Brazilian farmers in the tropics have maintained for decades their B. taurus x B. indicus hybrid dairy cattle population intermediate between both pure species. However, many small farms keep only one bull and do not practice artificial insemination (AI) (Table 1) and so to keep the herd composition intermediate they resort to periodical switching of the bull species, in a rather disorganized way. Rotational crossing of purebred sires is a better way to utilize heterosis (see below) but it requires either AI or controlled mating in farms with at least two bulls, and such farms are a minority (Table 1). Use of hybrid sires would be easier but less heterosis results. In the survey by Madalena et al (1997a) 40% of all 387 bulls were B. taurus/B. indicus hybrids. In that survey, 46% of the farmers intended in the near future to keep their herd hybrid, 13% intended to have a pure B. taurus and 1% a pure B. indicus herd, while 40% did not have or could not express a policy on this respect. Of those intending to keep a hybrid herd, 47% would use just one purebred bull, 43% a hybrid bull and 10% rotational crossing.  Experimental results on the relative merits of these alternative strategies are presented below.

Research on crossbreeding

Information on the performance of B. taurus x B. indicus crosses from three planned experiments are presented here in some detail as these studies are considered the most informative. Performance analyses at private or institutional herds are also considered. The EMBRAPA/FAO trial is discussed at length because of its unique factors that include, a) a carefully planned design, b) comparison of contemporary females of different genetic groups c) trial of animals in commercial herds, d) detailed recording, and e) recording for lifetime (12 year) performance.

The EMBRAPA/FAO trial on strategies for crossbreeding

Design, animals, farms and recording

Shortly after the EMBRAPA Dairy Cattle initiated its activities in 1975 (then named National Dairy Cattle Research Centre, CNPGL) a FAO/UNDP project was established to advise on animal breeding research and it became apparent that there was not consensus of technical opinion on what kind of cattle should be kept in the tropical areas. Some favoured intensification and use of  B. taurus and most recommended keeping B. taurus x B. indicus hybrids, although were not clear on how to do this (the majority advocated “fixing” ⅝ B. taurus:⅜ B. indicus at every individual farm, and some recommended crisscrossing). The early Brazilian literature (review by Madalena 1981) pointed to the existence of important heterosis for milk yield and reproduction, suggesting advantages of intermediate crosses, but there was insufficient experimental evidence in Brazil or in the international literature to provide recommendations on crossbreeding strategies. It was therefore decided to set up a trial to compare alternative strategies under the prevailing production circumstances of the Southeast Region, which then produced half the Country’s milk.

The strategies compared were 1) upgrading to Holstein, 2) developing a new breed from Holstein-Zebu foundations, 3) crisscrossing Holstein-Zebu (H-Z), 4) modified crisscrossing, with two generations of Holstein sires followed by one of Zebu sires (H-H-Z) and 5) continuous herd replacement with F₁ Holstein-Zebu heifers.

Female heifers (527) of six contemporary genetic groups of red and white Holstein-Friesian (H) x Guzerá (Gu) crosses were produced by AI at an experimental farm and distributed to co-operator farms where performance was recorded. The ca. 650 cows in the dam herd were hybrid dams available from a previous project plus purebred acquisitions. The six groups, referred to by their H grades (i.e. expected H gene fractions) were: ¼ (Gu x F₁, sire x dam breed), ½ (F₁ of H x Gu), ¾ (H x F₁), ⅞ (H x ¾), >31/32 (H x >15/16) and ⅝ inter se  (⅝ x ⅝). These groups were chosen to be close to those that would be generated by the crossbreeding strategies being examined. The number of H, Gu and ⅝ sires were 30, 15 and 8. These sires were basically unselected for milk yield (Madalena et al 1990a).

An essential feature of this trial was to have performance recorded under commercial management rather than at experimental stations, as the latter have bureaucratic problems which would make proper management difficult and they would not provide a realistic commercial environment. Also, the trial cost was halved and the likelihood of the trial being disrupted by administration changes were much lower (as unfortunately was justified later on). Co-operator private farms (65) in the main milk producing areas of the Southeast Region were chosen to cover a wide technological range, and were grouped for statistical analyses into two classes referred to as high and low management levels. Two experimental farms also participated, one of them, Santa Mônica, keeping 97 heifers, as it proved difficult to find co-operator farms with high management. Most data of the high management class came from this farm (i.e. 83% of the lactations). Climatic data and a map showing the location of the farms were presented (Madalena 1989).

With a few exceptions each farm received a cohort of six contemporary heifers, one of each crossbred group. The mean absolute age difference among the cohort was 36 d and the mean range was 83 d. Heifers were distributed to farms, at mean age 22 mo and mean weight 263 kg, where they were managed with no interference from the research staff. Performance was recorded up to the end of the last lactation after 12 y of age of the youngest cow in each cohort. Heifers were born between 1977 and 1981 and the trial ceased in 1994.

A decision was made to include only farms milking 2X with the stimulus of the calf, the prevailing practice. On recording days cows were fully milked with the stimulus of the calf as usual, except that no milk was intentionally left for the calves, nor were these calves allowed to suckle between milkings. Milk yield was recorded monthly (fortnightly at Santa Mônica) by local extension, dairy cooperative or EMBRAPA technicians and milk samples were sent to the research centre for protein and fat testing. The local recording was supervised by two research centre technicians and the Project research staff. The individual intake and ingredient composition of concentrate feeds were recorded along with milk yield. Cows fed a uniform ration quantity (the usual practice in the low management farms) were assigned the average herd ingredient consumption on the recording day. Further details were given (Madalena 1989, 1993a and Madalena et al 1990a,b).

Dairy traits

In the high management environment the >31/32 H had longer lactation length than the other groups (Table 2). Differences in milk yield between groups ½, ¾, ⅞ and >31/32 were small. The ¼ and ⅝ had very short lactation lengths and low milk yield. In the low management environment, the F₁ had the longest lactation length and highest milk yield; the ⅝ had very poor means for these traits and means of the other groups tended to decline as the Holstein-Friesian breed contribution deviated from ½.  Thus, the crossbred groups varied in their response to improved management, measured by the differential performance in the high and low management classes. Groups ¼ and ⅝  showed small response to improved management in milk yield and their lactation length was greatly reduced; groups ¾ and ⅞ had substantially increased milk yield with moderate changes of lactation length and the  >31/32 had large increments of both lactation length and milk yield (Table 2).  These results indicate that the nature of short lactations, a serious problem in tropical dairy cattle, differs between B. taurus, (unable to sustain lactation under stressful husbandry conditions) and high B. indicus grades, which showed very short lactations even on the better managed farms, i.e. they had a low genetic potential for this trait. An examination of causes of termination of lactation records in the high management group indicated that 89.5% of lactations in the ¼ terminated either naturally or by forced drying off because the limit of low yield was attained before the 2 mo pre-calving rest period. In groups ½, ¾, ⅞ and >31/32  the corresponding figure was between 36.6% and 44.8%, and in the ⅝ it was 72.4%. This suggested a more pronounced depression of milk yield due to pregnancy in the ¼ and ⅝. 

Higher B. indicus grades showed higher protein and fat content in both management groups and, consequently, the relative differences between crossbred groups for fat and protein yields differed from those for milk yield  (Table 2).  

Daily milking time (i.e. time elapsed between initiation and termination of milk removal at am + pm milkings), ease of hand milking and temperament scores (both assigned by the milkers) were recorded in a sub-sample of the project farms. The >31/32 and ⅞ crosses showed the highest milk flow (milk yield/milking time) but had intermediate milk yield. The F₁ and ¾ had the highest yield but intermediate flow, while the ¼ and ⅝ had the lowest yield and flow (Table 3). Thus, the crossbred groups with higher yields were not those with the higher rate of flow, which differs from the high positive correlation between these two traits usually found within B. taurus breeds. Madalena et al (1990b) showed that the extra benefit obtained from the higher yielding crosses outweighed their increased milking cost. Milk flow was higher for hand than for mechanical milking (suggesting over-milking in the latter) but the crossbred group x milking procedure interaction was not significant (P>0.05). The >31/32 were the easiest to milk by hand and the ¼ the hardest, the other groups receiving intermediate and similar milking ease scores (Table 3). The >31/32 were scored as the most docile and the ¼ as the least, and the other grades received scores slightly higher than 2 (docile) indicating that milkers did not considered cows of these crosses hard to deal with.

Genetic models

Dairy traits and calving interval were analyzed by least squares methods using two alternative models:

Y_ijk… =   b_o  + M_i + b_i1q_j+ b_i2z_j + farm_ik + other environmental factors + e_ijk…          Model [1]        

Y_ijk… = mean + M_i +G_j + MG_ij + farm_ik + other environmental factors + e_ijk…          Model [2]      

In model [1] the genetic differences among the six crossbred groups were described by the  regression of performance Y on two covariates; b₁ is the regression on the expected Holstein-Friesian gene proportion (q) and b₂ is the regression on breed heterozygosity z, defined as the expected proportion of loci in a particular cross that contain one allele from each breed. Thus, b₁ estimates the additive effect of breed substitution (Holstein-Friesian minus Guzerá), and b₂ estimates heterosis (Dickerson 1969). This model was proposed by Gardner and Eberhart (1966) and first applied to dairy cattle by Vencovsky et al (1970). Different regression coefficients were assumed for each management level (M_i, high or low) to accommodate the management by genotype interaction. In model [2] the genetic group class effect G_j (j=1,6) and the management by genetic group interaction (MG_ij) substituted for the additive and heterosis effects. The goodness of fit of Model [1] was assessed against Model [2].

An illustration for protein yield appears in Figure 3, where it may be seen that the lines predicting performance were not linear on the breed proportion q, due to the heterosis term. The different slopes in the two management groups reflect the GxM interaction. The additive breed effect (b­₁) was larger in the high management group while heterosis (b₂) was larger, in proportional terms, under low management. Model [1] fitted very well when the ⅝ inter se data were excluded from the analysis (as in Figure 3). This ⅝ group performance was well below that predicted under model [1] for all traits related to milk and component production (Madalena et al 1990a), which was interpreted as evidence of epistasis being broken up by recombination. This evidence is circumstantial, as genes are not identified to actually prove epistasis, but whatever the cause, the important practical result was the loss of performance in the inter se compared to other crosses.

The strength of crossbreeding genetic models rests on their predictive role, rather than providing a biological explanation of the results (Madalena 2001b). In this trial, breed heterozygosity z was confounded with the Gu fraction of the dams, so the heterosis found could theoretically have been due to maternal effects. But this does not reduce the model’s ability to predict the performance of purebred sire progeny, as was done to compare the proposed crossbreeding strategies (see below), since model [1] was duly validated against model [2].

Biases due to data handling

Although short lactations have long been recognized as a serious problem in tropical dairying (Rhoad 1935) two decades ago there was no consensus in the Brazilian literature on how to handle milk records. Three alternative procedures have been widely used: a) to adjust milk yield by the phenotypic regression on lactation length, b) to exclude from the analysis short lactations, considered “abnormal” and c) to use all records and not adjust for lactation length. The rationale for b) is that short lactations are presumed to be caused by environmental factors and so higher selection accuracy would be possible by disregarding them, particularly when reliable records on causes of terminating lactation are missing, as it is oftentimes the case. Procedure a) extends this reasoning to all variation in lactation length, which is completely removed, along with any associated variation in yield.

To examine this point, first lactation milk yield data of this project were separately analysed using each procedure, fitting models [1] and [2] (Madalena et al 1992b). Lactations shorter than 120 d were discarded in procedure b). The frequency of lactations <120 d duration and the lactation length were influenced by the genotype. As lactation length was highly correlated with milk yield (r = 0.85), removing variation in lactation length as in a) and b) reduced the estimated differences in yield between groups.  As an example, the mean lactation yield of the six crossbred groups in the low management level, relative to the F₁ mean, is presented in Table 4. The frequency of lactations shorter than 120 d was high in some groups, e.g. 20 and 26% in the ¼ and the ≥31/32 (hardly “abnormal” events). The crossbred group effects were attenuated in procedures a) and b) in comparison to c). As a consequence, the estimates of the breed difference and heterosis (b₁ and b₂ in model [1]) were reduced in procedure a) to 55 and 53 percent of the estimate under c), and in b) to 89 and 83 percent. Hence, procedure c) was used in the analyses of the EMBRAPA/FAO trial.

The reduction in the b₁ and b_2­ parameters was theoretically expected. Letting y and x respectively represent lactation milk yield and lactation length, the breed difference (b_1y*) and heterosis (b_2y*) parameters for lactation length adjusted yield (y^* = y – b_y.x­x_­) are b_1y* = b_1y (1- b_y.x b_1x/b_1y) and b_2y* = b_2y (1- b_y.x b_2x/b_2y) and therefore procedure a) yields a biased estimate of b_1y if b_1x > 0 and a biased estimate of b_2y if b_2x > 0, unless cov (y,x) = 0  (Madalena et al 1992b).  

Editing out abnormal lactations would also bias the comparisons, as the frequencies of abnormal lactations were also affected by genotype; the frequency of lactations terminated by injury, illness or death increasing with the proportion of H and the frequency of lactations terminated because of death of the calf decreasing with proportion of H  (Madalena et al 1989b).

Procedures a) and b) hide the effects of short lactations on yield. A thorough comparison of breeds and crosses should include all records, or, at least, report the frequencies of those discarded in each group. The same argument applies to other related traits, as for example, age at first calving, some times used to correct yield although it is also genetically influenced (Table 5). Note that there is no new principle involved in this reasoning and much has been written about the negative effects of selective recording in animal breeding, and also statistical textbooks warn about not correcting treatment effects for a co-variable that is itself affected by the main effect. Unfortunately, the lack of uniformity impairs the comparison of literature results obtained by different procedures, which is aggravated by incomplete reporting on data editing.

Madalena (1988) also found that, as expected, procedures a) and b) also reduced within breed selection efficiency. Facó et al (2009) confirmed that procedures a) and b) alter the crossbred group differences relative to c), although the residual heritabilities were slightly higher in b).   

Fertility, survival and herd life   

The performance of crossbreds in survival, fertility and herd life followed a similar pattern as in dairy traits, being superior in the F₁, decreasing with backcrossing and being very poor in the ⅝ inter se   (Table 5). Very few heifers (<4%) did not calve in the high management and differences among crossbreds were not important but differences were important in the low management.  There were important differences in age at first calving in both high and low managements with differences between crossbred groups being larger in the low management (Table 5). Teodoro et al (1984) reported similar results for age at puberty at the Santa Mônica (high level) farm where heifer weight gain was ≈ 0.4 kg/d. The longer lactation length and higher milk yield of >31/32 in the high management were accompanied by longer calving interval (CI, Table 5), which offset the advantages of this genotype for milk and component yield per day of CI. Conversely, the ⅝ had excellent reproductive performance in the high management but this was not compensated by their poor yield, so the milk yield/CI for this group was also very low. Important differences among grades were also found in herd life, following the pattern described above. Longer herd life implies a lower proportion of replacement heifers and therefore higher receipts from sales of heifers not needed for replacement. Cardoso et al (1999) estimated optimum economic herd life and voluntary culling to be 6.3 y and 10.7% respectively for the F₁ in a simulation of a system similar to the high management group.

The fate of cows entering the herd (distributed females that calved at least once) had three possible outcomes: died, sold for beef or sold for dairying. All cows were sold at the termination of the experiment, at 12 y of age (or more, if not dry at that time). As shown in Figure 4, for both management levels pooled together, mortality increased with the H fraction, being least for the ¼, which were mostly sold for beef, as were the ⅝, consistent with their low yield. It is impressive that most F₁ cows (63%) were sold for dairying at advanced age. F₁ age at first calving plus herd life summed 11.4 and 10.4 y in the high and low managements, respectively (Table 5).

Teodoro et al (1996a) showed that the salvage value of cows in this trial sold for beef (at mean age 8.1 y) decreased linearly with increasing H fraction, from a mean 921 kg milk equivalent in the ¼ to 665 kg in the ≥31/32. For cows sold for dairying, at a mean age of 11.8 y, only heterosis and not the breed additive difference was significant. Mean prices for the ¼, F₁, ⅝, ¾, ⅞ and  ≥31/32 were 921, 1093, 976, 1031, 929 and 408 milk equivalents.

Adult weight, height and growth traits   

Live weight and height of 263 cows were taken twice a year, in the rainy and dry seasons, in a sub-sample of 50 farms with easier access to portable scales (Madureira et al 2002).  Cows were 3 to 9 y old and had on average 5.6 weight and height records each. Quadratic curves fitted the age effects on weight, height and weight/height ratio (W/H), with maxima at 11.4, 8.3 and 12.4 y, respectively. No significant differences were detected among crossbred groups in the slopes of those curves (Madalena et al 2003). The mean cow weight over the 3 to 9 y  of age showed heterosis (Figure 5), as did the mean weight/height ratio, while the mean height decreased linearly with the H gene proportion (Madureira et al 2002). In contrast to milk yield traits, there was no evidence of epistatic recombination loss of performance in the 5/8 inter se cross. Martins et al (2004) also found important heterosis for weights of Holstein/Gir cows and Peroto et al (1997) also reported heavier adult weights in F₁ than in other Holstein/Guzerá and Gir crosses.

The weight/height ratio is related to body condition (e.g. Nelsen et al 1985), which may influence the nutrition cost because of the lower maintenance energy requirements for fat than for other tissues (e.g. Herd 1995). Reviewing Brazilian trials with fattening males Silva et al (2002) reported mean net energy requirements for maintenance in pure Zebu, ½ to ¾ H dairy crosses, and ≥⅞ H as, respectively, 71, 80 and 89 kcal/kg bodyweight^0.75._. The cost of feed for maintenance of cows (dry plus milking) accounted for about 55% of total feed costs in two herds averaging 3940 and 3300 kg/milk/lactation (Vercesi et al 2000, Martins et al 2003) and would be a higher proportion in herds achieving lower yields. A more accurate and detailed description of the crossbreeding effects on feed requirements is important to predict the production cost for each cross.

Birth weight and gestation length were described by Lemos et al (1984) and heifer growth by Flores et al (2004). In the artificial rearing phase no heterosis was found for average daily gain, which increased linearly with the H gene fraction. A poor adaptation to bucket feeding of calves with higher Gu fraction was observed. In the pasture rearing phase heterosis for daily gain resulted in a similar pattern as in Figure 5. Balieiro Neto et al (2007) also reported reduced rate of gain with increased H fraction in crossbred H x Gir heifers in pastures 

In a 112 d controlled feeding trial with 180 yearling heifers on two feeding levels (rainy season average daily weight gain 533g/d and dry season average 314 g/d) the F₁ gained faster, ate slightly more and showed better feed conversion ratio than the other groups in the rainy season. Increasing proportion of H decreased performance while heterosis was favourable.  In the dry season the same trends were similar but not statistically significant (Madalena et al 1992a, Paiva et al 1992). No significant differences (P>0.05) between genetic groups in dry matter digestibility were detected in either season. 

Resistance to parasites

Variation among cattle breeds in resistance to the tick Rhipicephalus (ex Boophilus) microplus was first described by Villares (1941). Lemos et al (1985) reported that tick burdens increased exponentially with the H fraction and the differences among grades expanded at higher levels of infestation (Figure 6). Similar results were found by Oliveira and Alencar (1990) who also reported that the number of torsalo fly (Dermatobia hominis L.) nodes per heifer were about 4 in grades of ⅝ H or lower and 7 to 8 in the higher H grades. Several studies were conducted comparing burdens of gastrointestinal worms, but the results did not show very clear genetic differences, perhaps due to the low natural challenge levels (Honer et al 1981, Paloschi 1981, Figueiredo et al 1985).

An intriguing result was obtained in an experiment run with the dams of this project’s heifers.  An estimate of the effect of ticks on whole lactation milk yield was obtained from the difference between cows artificially infested with R. microplus larvae and cows sprayed with acaricides. Tick burdens reduced the yield of the >15/16 H by 26% but did not affect the yield of the F₁ and ⅝ (Teodoro et al 1998). These cows had been assembled four years earlier from different farms so although they had since been kept together, crossbred group differences cannot definitely be attributed to genetic effects. Nevertheless the results point to the resistance to ticks as a component of heterosis for profit.

Brazilian work on molecular genetics of resistance to parasites was reviewed by Regitano et al (2006).

Economic evaluation

Profit per day (PPD = (income - expense)/current herd life), the chosen economic evaluation criterion, was calculated for each cow terminating lactations up to Aug 31 1985, when they had mean age 7.9 and 6.6 y in the high and low management farms. Note this was a partial evaluation as the trial continued after that date. However, cows staying in the herd were assigned an economic value according to their individual last lactation milk yield and regional prices.  Current herd life was the time interval from first calving to the end of the last lactation available.

Income = accumulated milk produced x milk price (based on fat test) + l0-d-old calves produced x price of calves + final cow value + value of cull heifer fraction associated to a heifer calving (i.e. the proportion of cull heifers per first calving heifer in the same grade x management level class). Dead animals had zero value.

Expense = concentrates + milking labour + milk transport + heifer cost up to first calving + cost of (cull + dead) heifer fraction associated to a heifer calving + all other costs, including roughages, pastures and fixed costs.

Income and expenses were based on individual performance or on crossbred group means in each management level, except for milk transport and “other costs”, considered equal for all crosses. Income and expense items were discounted to age 30 mo at 6% annual interest rate.  

All prices were expressed relative to farmgate milk price (milk equivalent, me). The initial (30 mo age) heifer price was assumed equal for all heifers. Further details were given (Madalena 1989). 

Average concentrate consumption of milking cows was 4.5 and 1.6 kg/cow/d in the high and low managements and profit was higher in the low group. After testing different genetic models a modification of Model [1] was used for separate analyses of PPD in each management level, fitting the breed difference and heterosis regressions (b₁ and b₂) with F₁ and backcrosses data (i.e., excluding the ⅝ inter se) (see Madalena et al 1990b for details).   

The results are shown in Figure 7, where only the predicted values are shown for clarity, as the model fitted the data well. The F₁ excelled in both managements, although the decline in profit with higher H fraction was more pronounced in the low management farms. In the high management group the ¼ had negative profit due to very low yield, while in the low management group the ≥31/32 showed negative profit, due to a combination of low yield and reduced fertility and health traits. The ⅝ inter se performed similarly to the 1/4, having negative profit in the high management and poor profit in the low management. 

It should be noted that the pasture and roughages cost was considered equal for all grades because, although this cost may be estimated from nutritional maintenance and production requirements, no precise description of those requirements were available for each cross. In Holstein x Brahman and Jersey x Brahman crosses, Solís et al (1988) reported lesser maintenance energy requirements per unit metabolic weight in each F₁ than in both parental breeds. The NRC (2000) indicates requirements in Zebus 30% lower than in European dairy breeds, and depending on whether heterosis for energy requirement is assumed or not the maintenance cost may be lower or equal in F₁ than in higher H grades (Madalena unpublished) and therefore the economic superiority of the F₁ might have been even larger had the maintenance costs been accounted for specifically for each cross. Kahi et al (1988) undertook a re-analysis of the present results considering equal energy requirements for all grades but unfortunately double counted the estimated costs of roughage/pasture and so obtained incorrect results.

Comparison of crossbreeding strategies 

After a few generations, Holstein-Zebu crisscrossing (H-Z) generates a herd composed by ⅓ H and ⅔ H animals and the H-H-Z rotation a herd of 3/7, 5/7 and 6/7 H (e.g. see Madalena 1981). Animals of such compositions were not present in the trial, but their performance can be interpolated from the regression model [1], which was shown to fit progeny of purebred sires data (i.e. F₁ and backcrosses). The new breed was represented by the ⅝ and, although a new breed should not necessarily be of that or any other strict gene fraction, nor indeed restricted to just two breeds (Madalena 1999), the ⅝ group provided a convenient source of advanced generations of inter se mating and also served to test the ⅝ “magic” fraction.  Upgrading to H was represented by the ≥31/32 and continuous herd replacement with F₁ by this group itself. Because the F₁ provided the highest profit per day, it was taken as the base to express the economic performance of the other strategies.

As may be seen in Table 6, in the high management the second best alternatives were upgrading to H and H-H-Z, while the H-Z would not perform well on account of the ⅓ H low milk yield. However, in the low management, the H-Z strategy was the second best, as the H-H-Z average profit dropped because of the 6/7 H low performance. Upgrading to HF would have disastrous consequences in the low management due to its low performance in most traits, such as milk and component yield, fertility, mortality and herd life. The new breed (⅝ H) profit, negative in the high management, was positive but still very poor in the low management.  The common strategy of small farmers of periodically switching the bull species generates a high proportion of animals with too high or too low H fraction to be as profitable as the intermediate crosses (Figure 7).

Sensitivity studies, performed by repeating calculations in 15 other situations of varying prices, indicated that the conclusions were fairly robust to major economic changes. However, tripling the milk fat bonus and rewarding milk protein, in line with the present day payment system of some processor firms, changed the H-H-Z profit under high management to 72% of the F₁ while upgrading to H was reduced to 57%, on account of its lower milk solids content (Madalena et al 1989c).  The fat bonus in the base situation was very low, 0.0415(fat % -3.3) (for milk with fat content >3.3% and 0 otherwise), so the results in Figure 7 and Table 6 likely would show little change in a situation of payment just for milk volume, disregarding milk solids content, as unfortunately is still the case in many regions.

It should be noted that the profit function reviewed here assumed that the cost of heifers of all crossbred types was identical, while in practice this may not be the case, so differences in profit between crosses should be weighed against their differences in heifer cost.

The above results showed that profit could be substantially increased, and losses avoided, by using an appropriate genotype in a given production system, without any additional changes in nutrition, health or other inputs. This, of course, does not imply that environmental factors should not be improved, but rather that both the genetic and environmental components should be considered in unison. The choice of germplasm is an integral element in the production system and it should be matched to the other available inputs. The view is sometimes expressed that there is no need to worry about genetics until the management is sufficiently improved to allow full expression of the existing available genetic potential. This view, however, fails to recognize that some genotypes have higher potential in a favourable environment but a lower potential in more stressful conditions. Therefore the notion that there is a genetic potential for each level of management is conceptually and practically more useful than an absolute potential (Falconer 1960). As illustrated by the large differences in profit shown in Table 6 ignoring genetic differences would be an unwise decision in any development programme.

Other results on crossbred group comparisons

No other studies involving planned comparisons of contemporary B. taurus x B. indicus grades have been conducted since the EMBRAPA/FAO trial but there have been several analyses of crosses within institutional or commercial farm records. These studies usually employed a statistical model that adjusts data for year, season and age of cow.  A study of a large number of lactations in the Holstein Breeders Association herds in Minas Gerais indicated a larger increase in milk yield with increasing H fraction (Table 7) than in the high management in the EMBRAPA/FAO trial (Table 2). In this study the possible confounding of herds and grades was carefully examined, but the milking method was not described and it is likely that most farms machine milked without the stimulus of the calf. Notwithstanding, the higher response of milk yield to improved management in higher H crosses has long been recognised in Brazil and abroad (Madalena et al 1990a).  Other reports by Grossi and Freitas (2002), Perotto et al (2010) and Santos et al (2011) in the States of São Paulo, Paraná and Ceara, respectively, confirmed the general trend observed by Freitas et al (2001) and the EMBRAPA/FAO trial, except that Santos et al (2011) did not find higher yields in grades above 60% H in cows yielding about 6000 kg milk/lactation with average temperature 25^oC and relative humidity 70%.

However, the conclusion that increasing inputs allows higher yields in high B. taurus grades is only part of the story; as costs should also be considered, and profit, not just milk yield, should be the appropriate basis to make decisions on breed use. Unfortunately estimates of profit are not available to compare crosses other than those of the EMBRAPA/FAO trial discussed above but a related trait, lactation milk yield/calving interval (MY/CI) was reported in some studies. . It may be shown that yield per day of calving interval is more related to profit than yield alone, as it considers also the dry cows in the herd. Data on MY/CI was collated from seven studies and an additive-heterosis model was fitted to the grade means in each report weighted by their standard error or number of observations. Results are shown in Figure 8, which also details the sources of data. Other reports were disregarded due to one or more of the following reasons: confounding of herds and genetic groups; yield was adjusted for lactation length or vice versa; unknown mixtures of Holstein, Jersey or other breeds were grouped together under “European”; there was no distinction of whether a cross was a backcross or inter se; there were strong data edits (e.g. deleting lactations <200 d duration); some crosses were milked with the stimulus of the calf and others without it; the milking system was not described; there were too few animals per group or flawed calculations in the analyses. It is felt that such analyses just blur interpretation of results.

In Figure 8, up to about 9 kg milk/d the EMBRAPA/FAO results in red and white H x Guzerá crosses (lines 1 and 6) parallel those in the more popular black and white H x Gir when also manual milking without the calf (lines 4 and 5). However, at higher feeding levels with mechanical milking without the calf (line 3), H x Gir performance improved with higher H grade. In this study, Ferreira and Ferreira (1998) reported that 62% of culls of F₁ cows occurred in the first lactation, mostly because of lack of adaptation to mechanical milking without the calf.

Two studies comparing crossbreeding strategies are summarized in Table 8, corresponding to lines 4 and 3 in Figure 8. At the EMBRAPA farm the new breed was represented by the progeny of the two top milk yield sires available in a multi-breed hybrid sire progeny testing programme (Madalena 1999), suggesting that one generation of selection for milk yield was not enough to counteract the heterosis loss of performance and come close to the F₁ performance. At the EPAMIG farm 8 non tested sires of the same programme were used and it was shown that their progeny were more strongly culled for low yield and short lactations than the H-H-Z. In a dry, hot, climate Barbosa et al (2008) reported 7.2 kg daily milk yield (adjusted for lactation length) in H x Gir F₁ and 4.7 kg in ⅝ inter se. Corresponding fat yields were 0.313 and 0.193 kg/day. This loss of performance in the ⅝ inter se is very similar to that observed in the ⅝ inter se in the EMBRAPA/FAO trial (Table 2). A reduction in milk and fat yield and fertility following inter se mating was also reported when comparing the first vs further generations in the Pitangueiras breed (⅝ Red Poll:⅜ Guzerá) (Lôbo et al 1979a,b, 1988).

Martinez et al (1988) reported similar milk yield per day of calving interval in European/Zebu F₁ and ⅝, but as the parent genetic composition was not available they assigned an arbitrary value to heterozygosity in this group. Moreover their data were not appropriated for comparisons between crosses due to grouping together different European breeds, grouping different zebu breeds, confounding of breeds, crosses and farms and editing out short lactations and long calving intervals.

Making use of the Girolando Breeders Association records, Facó et al (2002) analyzed lactation milk yield separately in three feeding systems finding non significant trends between grades in the “extensive” feeding (P>0.05) while in the “semi-intensive” and “intensive” systems yield increased with the H fraction, in line with the results of Table 2. They also presented results for ⅝ cows but assigned an arbitrary value to heterozygosity in this group. When Model 1 was fitted the estimates of breed additive effects and heterosis differed substantially between systems, with heterosis being less important in the “intensive” feeding, as observed in other studies reported here. In a later analysis on further data (Facó et al 2008) the three feeding systems classes were replaced by herd-season effects and milk yield per lactation and per day of calving interval strongly increased with higher H fraction. Heterosis and recombination loss of epistasis were also observed. Utilizing the same data source Camilo et al (2009) reported similar grade trends and Freitas et al (2006) using a different genetic model, including maternal effects but not recombination loss of epistasis, reported significant but lower breed additive effects and direct (individual) heterosis and also found maternal heterosis. However, given the crosses available in this dataset, the individual and maternal breed proportions, heterozygosity and recombination loss of epistasis values are correlated. Hence, changes in the multiple regression genetic models are expected to change the regression coefficients and interpretation of this data. T heterosis. Given the crossbred groups involved in this data bank the breed additive, heterozigosity and recombination coefficients are correlated, so changes in the model are expected to alter the parameter estimates. The model goodness of fit was not tested in either study. It would appear, however, that the replacement of the system stratification by herd-year effects absorbed into the genetic group effects the system x group interaction. A comparison within production systems would provide a clearer evaluation of the genetic group differences. The milking system would be an obvious element of management classification, particularly since Facó et al (2007) reported higher environmental variance in the F₁ than in ¾ and ⅞ which might have been partially due to the pronounced effect of milking system on the former group. Curiously, although ⅝ is the desired breed composition of the Girolando, and this dataset is the obvious source to compare the joint effects of crossing and selection in the ⅝ inter se group vs other crosses, such analysis has not yet been conducted. 

The EPAMIG trial on maternal B. indicus breeds

The Agricultural and Livestock Research Organization of Minas Gerais State (EPAMIG) compared Gir, Guzerá and Nelore as maternal breeds to obtain F₁ crosses with Holstein, as part of its research programme on low-cost pasture-based dairy production (Ruas et al, 2008). The trial was conducted at the experimental farm located in Felixlândia, 18^o7’S and 44^o8’W, in a savanna area with an Aw climate (Koeppen classification) and annual rainfall 1126 mm. The herd originated partly from EPAMIG farms and partly was purchased. During the rainy season the milking cows were kept in rotationally managed pastures of mainly Brachiaria decumbens and B. brizanta and in the dry season were supplemented with maize silage and sugar cane-urea. Minerals and concentrates were fed regularly. Cows were milked 2X in a walk-through tandem parlour with the stimulus of the calves during the whole lactation. During the first two months of lactation either am or pm, calves suckled one teat during milking, when the corresponding teat cup was blocked, and after two months calves still suckled a few jets before milking to stimulate milk let down and suckled the residual milk after the pm milking for about 30 min. This procedure promoted a complete and calm milking. Mastitis tests were performed monthly. First calving heifers were calmly familiarized with the milking routine. Natural service was practised with bulls of proven fertility in a 1 to 50 cow ratio. Standard vaccinations against foot and mouth disease, rabies, blackleg and brucellosis were practised. Females were annually tested for brucellosis and TB. Calves were chemically de-wormed and ticks were controlled by spraying as necessary. This system was designed to provide simple, low-cost practices fitted to H x Zebu F₁ animals, and made possible by them. Note that this was not a primitive system, on the contrary it incorporated many technical elements of the various fields involved and it was economically very competitive, as net profit was 29% of the total receipts (Moraes et al 2004). Another well documented dairy system based on restricted-suckled H x Guzerá F₁ in irrigated pastures was described by Alvares et al (2001), who reported for the first year of the project net profit equal to 36% of the capital invested.

The results, summarised in Table 9, show that the H x Gir F₁ had the longest lactations and the highest milk yield per lactation and per day of calving interval. The H x Nelore F₁ ranked the lowest in these traits although showing shorter calving interval and earlier ages at first service and at first calving. The H x Guzerá F₁ had intermediate performance between the other two genotypes in all traits except weight, in which it equalled the H x Nelore, both being about 4% heavier than the H x Gir. In this herd there were no differences in calf weaning weight among dam genotypes.

A whole research programme is conducted by EPAMIG aiming at fine tuning the F₁ based production system. Ruas et al (2011) reported that heifers trained to be handled at milking showed 3180 kg milk yield in the first two lactations, a 15% increase over non-trained animals. Several other production system options were/are being studied, including the effects of fixed-time AI protocols, use of X enriched semen, MOET and IVF, production of F₁ with natural mating, calf and heifer rearing, weight at first service and daily milking frequency (e.g. EPAMIG 2010).  

Three-way crosses

An experiment comparing the lifetime performance of the progeny of Jersey, Brown Swiss or H sires crossed to ½, ⅝ and ¾ H x Gir dams was run at EMBRAPA’s Santa Mônica Experimental farm in a machine-milking artificial calf rearing system. Albeit based on small numbers of animals, the results (Table 10) were in line with the international literature (references in Teodoro and Madalena 2002a,b). Longer herd life, lower weight (lower maintenance cost), lower age at first calving and higher fertility resulted in higher profit of the Jersey crosses; their advantage increasing when payment was for milk solids because of their higher milk fat content. The Brown Swiss crosses had the poorest performance. The conclusion was that in systems of artificial female calf rearing and slaughtering of male bobby calves the Jersey crosses appear to offer important economic benefits to farmers, which would be even greater if payment on milk protein and fat was implemented (Teodoro and Madalena 2005).

Beef traits in dairy crosses

In five fixed-time or fixed-weight experiments reviewed by Barbosa (2001) H x Nelore or H x undefined B. indicus F₁ males under confined conditions showed an average daily gain 1.1 kg/d, 22% higher than Nelore and similar to or higher than several B. taurus beef breeds x Nelore crosses. Razook et al (1986) reported that F₁ H x Nelore and Brown Swiss x Nelore males grew faster than Nelore and several beef crosses but had lower dressing percentage yield and subcutaneous fat thickness than the Nelore at the same age and lower dressing percentage than the Canchim x Nelore (Table 11). 

Madalena et al (1989d) reported that pasture-raised bullocks that were 0.56 to 0.69 dairy B. taurus x B. indicus crosses weighed 429 kg at age 27 mo, only 3% less than ¼ Chianina/ ¾ Nelore under comparable conditions.

Alves et al (2004) reported estimated percentages of muscle, fat and bone in carcasses of F₁ H x Gir, H x Guzerá and Indubrasil, all slaughtered at a fixed weight of 450 kg, of 57, 27 and 17;  59, 28 and 15 and 53, 34 and 15%. Santiago (2004), based on a large number of observations, estimated that Nelore x Gir steers took 6 mo more to attain the required commercial finish than the Nelore. Jorge et al (1999) did not find differences in the fat and protein composition of serially slaughtered bulls of Gir, Guzerá, Nelore and Tabapuã breeds.

The Holstein and Brown Swiss are known to have high growth rate and carcasses with high muscle and bone percentages and low subcutaneous fat, in comparison with beef breeds, which was consistent with the above results. As shown above, in the predominant tropical Brazilian dairy systems, sale of surplus progeny and culled animals for beef provides a significant proportion of farmer income and despite of their lower carcass yield and quality dairy crosses are being used for beef production and play a role in the production system as a whole. Indeed, it has been suggested, both by EPAMIG and by the Minas Gerais State Extension Agency, that the F₁ dairy system might be used as a terminal cross to Nelore, to produce beef calves, under the slogan dairy cow/beef calf (e.g. Almeida and Pires 2004, Marcatti Neto et al 2004), because grades ≥¾ H are not considered suitable for pasture production due to poor adaptation.

Practical developments

Crossbreeding in practice

As indicated by the surveys described above, most dairy farmers wish to maintain their herd intermediary between B. taurus and B. indicus. Several strategies may be used to that end, which make use of different proportions of profit heterosis and differ also in their practical implementation difficulties. Rotational crossing was the second best strategy predicted from the EMBRAPA/FAO trial, but this alternative is feasible only for the minority controlling reproduction, and in addition, without artificial insemination, keeping two bulls would not be economic for small farms. The common strategy of small farmers, still widely practiced today in Brazil, of periodically switching the bull species, is easy to implement but is not rewarding because of the high proportion of less profitable animals with too high or too low B. taurus fraction, as discussed above. Using hybrid bulls of a new composite breed is also easy to implement for the dairy farmer but the heterosis break down in the inter se crosses would reduce profit. The same would apply to the recurrent use of F₁ bulls proposed by Hickman (1979). Selection may increase performance in a new breed but it is unlikely that selection for fertility and health could be effective unless a very large population of recorded animals was available, which is not the case in Brazil. Thus, because of the shortcomings of other alternatives to maintain crossbred herds, continuous replacement with F­₁ appears then as a convenient strategy to capitalize on their economic superiority. This strategy requires the supply of F­₁ replacements, because most dairy farmers can not keep the pure breeds to produce the F₁.  Profitability for a dairy farmer will depend on the relative price of different sources of replacement heifers vs the achieved net profits of the different genotypes of heifer. As an example, in the low management results of the EMBRAPA/FAO trial, the total net profit of an F₁ heifer over its herd life (7.1 y, Table 5) was equivalent to the price of 11921 kg of milk (7.1x365x4.6, Table 6), and its superiority over an average heifer of the H-Z rotation, the second best alternative, was 41% (Table 6), equivalent to 4888 kg of milk, which is the break even point to choose buying the F₁ (Madalena 1993a). In this context, the supply of F₁ replacements may be seen as just another input for the dairy farm, subjected to the same market forces and to the general problems of modern agricultural organization, which often opposes farmers and input suppliers, yet some equilibrium price is achieved and extra farm inputs are increasingly used. Buying genetically superior female replacements is common in the swine and  chicken industries and there is no reason why the same could not be applied in cattle, provided it was economically warranted.

A simple scheme of ranch production of F₁ heifers was suggested based on existing commercial practices (Madalena, 1993b). It was calculated that a 1000 F₁ cow dairy herd could be maintained by replacements from a 600 B. indicus cow ranch, 60% of which was inseminated with B. taurus to produce F₁ heifers for sale and the other 40% mated to B. indicus to obtain the replacement heifers for the ranch herd (based on 80% weaning rate). A simulation study for a typical cow-calf ranch indicated that keeping a Gir herd to produce Holstein F₁ heifers had much higher profitability than a conventional Nelore beef herd, mainly due to the higher sale prices of the F₁ heifers (Guimarães et al 2006).

Two large ranch operations specializing on the production of H x B indicus F₁ by AI were described by Fiúza (2001) (a 1100 Gir dam herd for 40 y) and by Ferreira et al (2001) (770 Indubrasil or zebu crossbred dams). A survey in Minas Gerais identified 267 farms with more than 42,000 cows producing F­₁ heifers for sale, which under conservative assumptions could supply some 1.5% of the replacement females needed annually for the whole State. Sixty eight of those farmers replied to a mailed questionnaire. These had a total of ca. 16,000 B. indicus cows, 62% of which were inseminated/mated to B. taurus to produce F₁ and the rest to B. indicus. Sixty two percent of the F₁ were produced from Gir-type dams. The frequencies of farms using AI, natural service or both for crossing were 55, 28 and 17% respectively. A range of farm types was apparent in this sample, from large ranches selling all the F₁ females and keeping most F₁ males for fattening, to smaller herds keeping most F₁ females for dairying and selling a large proportion of males as weaners. Half of the F₁ producing farms were located in beef cattle regions and sold females to the dairy regions, a form of stratified crossing, where the replacement female is raised in the less expensive regions. Prices of F₁ females were considerably higher than prices of other crosses, indicating that buyers expect a higher profit from those animals. Almost all (99%) F₁ producers intended to continue this activity, indicating as the main reasons profitability, easy marketing, milk yield and satisfactory growth/fattening of males (Madalena et al 1997b).

Despite the profitability of specialised production of F₁ by AI this sector has not grown substantially. Santiago (2004) identified the shortage of dairy B. indicus cows and lack of management abilities, access to capital and the tradition of beef cattle producers as limiting the adoption of dairy F₁ production. The advent of synchronised AI, increasingly adopted in Brazil, should facilitate F1 production, overcoming some of the difficulty of AI in extensive cattle production systems. But the shortage of dairy B. indicus breeds is difficult to overcome because the high prices of these pure breeds (see below) mean it is more profitable to become a Gir or Guzerá pure breeder. Santiago (2008) described a 3000 cow project to develop a Guzerá-Nelore-Gir (GNG) composite from commonly available beef Nelore foundation animals inseminated/mated with dairy Gir and Guzerá proven sires or young bulls being sampled. The GNG dams inseminated with H produce the “Calciolândia” F₁ (named after the farm on which it was first developed).

A good market also exists for the ¾ H, which usually do not need the calf stimulus for milk let down and are then preferred by farmers opting for artificial calf rearing. Thus, some farms keep F₁ H x Gir inseminated with H and sell the ¾ heifers, buying in F₁ replacements (e.g. Machado et al 2002).  A large farm, Santa Luzia, in Minas Gerais, producing more than 15,000 kg milk/d in a rotational pasture based system is a practical example of large scale application of such crossing (Coelho 2004). The farm keeps H x Gir F₁, ¾ and some ⅞ cows. The latter two groups are obtained by AI at the farm and the F₁ replacements come from another farm keeping a Gir herd in Pará State. Most ⅞ and some ¾ females are sold for dairying. Female calves of all groups are artificially reared and male calves of all groups restricted suckle the F₁ cows, which are trained to adopt the sons of other dams, and are milked with the calf stimulus. Males are fattened in confinement. Vercesi et al (2011) reported higher milk yield and CI with increasing H fraction in this herd; mean milk yields per day CI in F₁, ¾, ⅞ and 15/16 H cows were 7.8, 9.7, 9.9 and 9.6 kg/d. The milk suckled from the F₁ by the male calves was not accounted for and probably would mean that the F1 produced similar or higher milk yield than other crosses. At any rate, it is the profitability of the system as a whole and not that of the individual H grades that is important for such scheme. This farm is currently moving to using oxytocin injections to milk the F₁ without the calf, a not wholly tested practice.

Applications of biotechnology

Sexed semen (semen sorted based on X-chromosome content) has been available for some years and an expression was presented to calculate the net present value of a sexed semen dose (Madalena and Junqueira 2004). The difference in value between male and female progeny determines the value of using sexed semen given that sexed semen is both more expensive and generally less fertile than unsexed semen. The value of sexed semen was estimated to be greater for a specialized H dairy farm producing high priced Gir x H F₁ as excess females than for a farm producing the reciprocal F₁ cross from Gir dams and selling males at weaning, because the difference in value between sex of progeny is largest when bobby calves are slaughtered. For the same reason sexed semen is not valuable to dairy B. indicus pure breeders who sell males and females for similar prices.

There were early suggestions to use MOET (multiple ovulation and embryo transfer) to produce B. taurus x B. indicus F₁ crosses  (e.g. Charles Smith (†) personal communication 1990). Calculations indicated that it might be economically feasible although very sensitive to the technical efficiency (Madalena 1993a, Teodoro et al 1996b). In 1998 an international AI firm announced that it would sell in Brazil embryos produced from H ovaries recovered from USA abattoirs fertilized with Gir semen, but this never materialized. However, as described by Viana et al (2010) the Brazilian embryo industry grew dramatically in the last decade. The number of embryos produced increased from ca. 72,000 in 2000 to ca. 290,000 in 2009, 60,000 of which were from dairy breeds, and 230,000 from B. indicus. In the latter, 96% of the embryos were produced by in vitro fertilisation of oocytes (IVF) vs 27% in B. taurus. In 2008 Brazilian companies accounted for 9% of the global total of in vivo produced embryos but 67% of those produced by IVF. The low efficiency of cryopreservation of B. indicus embryos means that less than 6% are frozen. The authors attributed the predominant use of in vitro embryo production in B. indicus to their high oocyte production and quality leading to 2.7 pregnancies/donor/OPU (ovum pick-up session), compared to less than 1 pregnancy/donor/OPU in B. taurus. Use of sexed semen with IVF is particularly attractive because very little semen is required and the disadvantages of higher cost and lower fertility of sexed semen when used in AI all but disappear when used with IVF.

A large scale application of IVF with sexed semen to dairy cattle breeding was reported by Pontes et al (2010). Gir (617), ¼ H x Gir (44), ½ H x Gir (37) and H (180) donor cows were used for IVF embryo production in a total of 5,400 OPU sessions. These generated 90,000 oocytes, which were matured and inseminated with frozen-thawed sexed semen of 8 Gir (on H oocytes) and 7 H sires (on the other breeds) and then air-shipped fresh in microtubes with culture medium, in special incubators, to eight farms in Pará and Mato Grosso, up to 2000 km away. There they were transferred into recipient females following a synchronised embryo transfer protocol. They reported mean oocytes recovered and mean pregnancies/donor/OPU in Gir (17.1, 1.2), ¼ H (20.4, 1.3), ½ H (31.4, 1.7) and H (11.4, 0.7), indicating strong heterosis for oocyte and pregnancy yields for IVF based on oocyte rescue. The overall pregnancy rate was 39% and 91% of progeny were female for H and 87% for Gir sires. Considering the large number of embryos transferred and the unusual logistics for transporting embryos over long distances the authors concluded that the results were “commercially acceptable”. 

The embryo industry is currently geared to the production of elite animals and prices may remain too high for widespread application in commercial dairy production, but prices might continue to fall, driven by increasing competition and use on a larger scale.  As an example one of the largest dairy processors in Brazil, a cooperative based in Minas Gerais,  offers to its member farmers three plans for, a) buying F₁ H heifers pregnant with ¾ H or F₁ female embryos, or b) sending their own recipient heifers to receive the oocyte at the embryo transfer center, or c) to have the embryos transferred at their farm, with fees equivalent to 5660, 2590 and 1020 kg milk, respectively. The embryo procedures are contracted to specialized firms. Special payment installments at subsidized interest rates are available (Estado de Minas, 2011). The difference in fees in a) minus b) provide an estimate of the price paid by farmers for the F₁ recipient heifer above its own heifer, which is equivalent to 3070 kg milk and well within the 4888 break even point discussed above. While none of these calculations are precise, they should, nevertheless, provide a good approximation for sound decisions.  

Given the commercial rates of pregnancies/OPU, as summarized above, other sustainable systems to produce F₁ might be devised, such as oocyte rescue in heifer rearing farms or use of dairy recipient cows in efficient well organized farms selling excess heifers. However, an attempt to transfer F₁ embryos on standard dairy farms gave poor results, with a pregnancy rate of only 23%, compared to 40% in the experimental centre (Goulart et al 2009).

Progeny testing

Conventional small scale progeny testing programmes, including recording, are run by the respective breeders associations of Gir, Guzerá and Girolando jointly with EMBRAPA, which provides genetic evaluations using BLUP-REML techniques. Genetic evaluation includes both purebreds and crossbreds and the B. taurus fraction is statistically adjusted using fixed effects for grade. The bulls belong to the individual breeders and AI companies contract semen production with them.

Progeny testing of dairy bulls was first initiated in Brazil in 1976 by an EMBRAPA∕FAO project on ”Development of the Brazilian Milking Hybrid (Desenvolvimento do Mestiço Leiteiro Brasileiro, MLB)” aiming at developing selection criteria in multi-breed B. taurus x B. indicus crossbred populations. The main research results were the low correlations found between milk yield and tick resistance, the negative  correlations between milk yield and body  weight, the high frequency (33%) of hybrid bulls not suitable for semen freezing and the demonstration that a fixed grade was not necessary for breed development  (Madalena 1999). After sampling 121 young sires the project was terminated in the 1990s due to loss of support from new EMBRAPA authorities. However, the in-service trained personnel and the logistic structure served as a base for the further progeny testing programmes established in other breeds.

The Gir progeny testing programme has been run by ABCGIL and EMBRAPA since 1985, based on guidelines developed by the EMBRAPA/FAO project. There are some 40 pure breed studs providing young sires for sampling. Initially only six to ten sires per year could be sampled but this number grew to the present 32 as the commercial success generated enthusiasm in the breeders to sample more sires, and they met the corresponding costs. In the last three years summaries of 16 to 19 proven sires were published, tested on average 41 daughters and 8 different herds each (136 herds total). About 70% of the daughters are B. taurus crossbreds and 30% Gir purebreds. Traits evaluated include milk, fat, protein and lactose yields and contents; several conventional conformation measurements, including sheath length; milking facility and temperament scores and κ-casein and β-lactoglobulin genotypes. Up to 2010, 203 sires have been sampled. The results have recently been reviewed by Verneque et al (2008) and Vercesi et al (2010) who reported the genetic progress in 305-d milk yield to be 1% of the mean per annum. More recently a separate small-scale programme was established by another breeders association, ASSOGIL, jointly with EMBRAPA but no estimates of genetic progress are available.  The progeny test creates a high commercial value for sires highly ranked on milk yield, causing the usual genetic bottleneck due to heavy use of a  few ancestors (Vercesi et al 2007). Reis et al (2010) estimated an effective population size of N_e = 67 in the Dairy Gir herds and noted an important decline in N_e after the first sire summary was published.

The beef and dual purpose Guzerá have separate evaluation programmes. Although a formal selection criteria is not defined, the dual purpose programme emphasizes dairy and growth traits evaluated under “realistic” production circumstances of little concentrate supplementation (Penna et al  2005). Three sources of information are included in a single data bank for genetic evaluation: a conventional progeny testing, a MOET scheme of highly selected donor cows to produce half-sib families and dairy records of cows not belonging to any of these two sources (Penna et al  2002). The MOET nucleus now evolved to mostly OPU-IVF and is an open scheme designed to obtain 24 full-sib families per year, composed of at least 4 females and one male full-sibs, by hierarchical insemination with 4 bulls (V.M. Penna, personal communication). Initially the aim was to obtain 12 families, which was doubled due to the resulting commercial interest. A single sire summary is published including all programmes. Up to 2010 299 sires were summarized (74 sampled sires and 100 full-sib families). Further details were given by Vercesi et al (2010). For cows in the dual purpose herds Peixoto et al (2006) reported genetic progress in 305-d milk yield to be 0.3% of the mean per annum. A genetic lift of 12% was observed in the first four years in the embryo transfer nucleus herd, which then stabilised, perhaps due to the special care taken to maintain inbreeding low and to selection for other traits. Peixoto et al (2012) reported N_e= 98 for animals in the dairy improvement programme with 5 generations of ancestors known.

The Girolando progeny testing programme initiated in 1997 and has sampled 48 sires plus 57 sires currently in the pipeline. The sire summaries include the Predicted Transmitting Abilities (PTA) for dairy and type traits and genotyping for DGAT1 and four disease markers (Silva et al 2011).

A Holstein sire summary (not a progeny test programme) is prepared by EMBRAPA using the breeders association records, which are mostly from the South Region. In 2011 the data bank for milk yield included 2,426 sires and 144,000 cows in 2,148 herds (Costa et al 2011). The summary included 87 sires, 17 of which were born in Brazil. PTAs for milk, fat and protein yield and conformation traits (STAs) were reported as well as genotypes for the same disease markers as in the Girolando. Genetic gain in milk and component yield has been disappointingly low in Brazilian Holsteins. In animals born from 1984 to 1998 in Rio Grande do Sul, Boligon et al (2005) found genetic trends of 0.14 and 0.05% of the mean for milk and fat yield, despite of h² estimates of 0.30 and 0.28, respectively, and in the national databank in animals born from 1968 to 1996, Costa (2005) reported genetic trends to be lower than 0.2% of the means for milk and fat yield, even though h² were 0.25 and 0.22 and the genetic correlations between the traits in Brazil and the USA were 0.85 for milk and 0.88 for fat yield (Costa et al 2000). Imported Holstein semen accounts for a large proportion of semen sales of this breed (e.g. 92% in 2009 to 2011, ASBIA 2012). Costa (2005) stated that there are no selection programmes designed to suit neither national objectives nor recommendations for the better use of imported semen, which is usually based on evaluations conducted in the exporter countries via marketing of the commercial representatives and subsidiary companies based in Brazil. One cause of the low genetic progress may be the weak selection pressure for yield traits, as prices of imported semen are less related to yield traits than to the relationship with famous ancestors (Madalena et al 1985). The mean inbreeding coefficient of 392 Holstein sires available in Brazil was 5.1% and in 92 Jersey 6.5% (Soares et al 2011). The global decay in fertility and health of the Holstein breed has been documented also in Brazil and has been transmitted into the crossbred B. indicus composites (see Madalena 2008). Resorting to other breeds seems appropriated and needs local research, e.g. Junqueira et al (2010) reported that the introduction of New Zealand Friesian in a H x Gir commercial herd improved fertility and milk solids content and reduced milk yield.

Concluding remarks

• The economic efficiency of production systems depends on an appropriate combination of land, labour and capital determined by their relative prices in each country and their availability in each particular situation, thus generating variation in farming practices. The use of high levels of concentrate feeds and other inputs in the dairy systems of North America and Europe are justified and made possible because of the high support prices received. On the other hand, the production strategy with tropical pastures relies on the exploitation of the ample solar energy for the efficient photosynthesis of fast-growing C₄ grasses and of water availability. Pasture quality is not high and concentrates are relatively expensive, needed more appropriately to feed humans or monogastric animals. However, very high stocking rates are possible. Adapted cattle may graze tropical pastures without need for expensive buildings or equipment to control the environment or high levels of chemical control of parasites. Thus, tropical dairy systems may still be very competitive, locally and globally. Pasture-based systems also have less environmental impact than confinement-based systems (e.g. Rotz et al 2009). The notion inherited from intensive temperate dairy systems that very high production per cow is desirable and is required for high profit has been harmful to the dairy industry progress in Brazil (and elsewhere).

• In a country as large as Brazil, with many geographical differences and wide variation in income, several production systems will co-exist. Given the existence of strong genetic by environment interactions, specific breeding strategies are required to match each production system inputs and constraints.

• Most of the results that have been collected in Brazil come from the Southeast region, where climate is less stressful than in other regions of Brazil, and even in the low management farms that were examined, twice-daily milking was practised. It is not known to what extent the results may apply to the many family farms milking once-a-day and to farms operating under more stressful environments.

• Perhaps the main conclusion from the results presented is the overwhelming importance of heterosis for profitability, resulting from cumulate effect of heterosis for most traits. Crossbreeding strategies to capitalize on heterosis have been tested and each alternative has its own benefit, costs and feasibility. Continuous replacement with F₁ maximizes heterosis but requires special organization. Rotational crossing comes second, but requires use of AI or controlled mating in farms large enough to justify keeping more than one bull. A hybrid breed may be used with natural mating but would suffer from recombination losses. In vitro fertilization with sexed semen of OPU oocytes and synchronised embryo transfer is already commercially practiced and likely will grow, facilitating the production of F₁ and ¾ replacement heifers.

• Rutledge (2001) noted that in B. taurus x B.indicus crosses the F₁ has generally been superior in milk yield and reproduction and the F₂ failed to meet expectations. He analyzed data from 80 world literature reports on dairy crosses and found evidence of breakdown of favourable epistatic effects for most traits and concluded that the 100-year efforts to breed a tropically adapted dairy cow from  B. taurus x B. indicus foundations had failed because recombination of incompatible genetic interaction systems caused too great a load on the advanced crossbred generations. He suggested the use of in vitro fertilization and allied technologies to exploit the obvious advantage of the F₁, much in agreement with the results presented herein. Nonetheless, it must be recognized that not all farmers will be able or willing to buy F₁s and some may prefer the simplicity of using a hybrid bull, although the combination of large scale investment and managerial/commercial organization required for optimal results in the development of tropical dairy breeds is not yet available (Madalena 2002).

• A current compilation of over 130 studies of dairy crossbreeding in low input tropical systems globally found that average lactation yield across studies substantially overlaps with the Brazilian results reviewed here (Gibson and Jeyaruban, personal communication). Hence the results on performance of different breeds and crossbreeds from more than 40 years of research in Brazil are expected to be relevant to other relatively low input tropical dairy systems. The results on use of technologies such as AI and IVF and sexed semen, however, will need more careful evaluation before extrapolating to other countries, as Brazil has many economic, structural, physical and human capacity advantages that may not be available in some other countries. Nevertheless, there could well be good opportunity to apply the range of technologies to both establish and subsequently support tropical dairy industry development, including small-holder based dairy systems elsewhere.

• Unfortunately, the results do not provide direct comparison between Brazilian B. indicus dairy or dual purpose breeds and indigenous breeds elsewhere. However, some Brazilian breeds have been improved from their originally imported base populations and may well be superior for tropical dairy production than indigenous breeds in many countries and it would be desirable to properly test Brazilian tropically adapted dairy breeds against a range of other promising breeds globally.

References

Almeida E F L and Pires J A A 2004 A utilização da vaca F1: visão da EMATER-MG. In Anais do 5⁰ Encontro de Produtores de gado Leiteiro F₁ FEPMVZ Editora, Belo Horizonte, p. 169-184 http://www.fernandomadalena.com/site_arquivos/870.pdf 

Alvares J A S, Holanda Jr E V, Melo M V M and Madalena F E 2001 Produção de leite em pastagens tropicais irrigadas: uma alternativa econômica, In: Produção de Leite e Sociedade Cap. 18. Madalena F E, Matos L L and Holanda E V (Eds.) FEPMVZ Editora, Belo Horizonte, p. 275-294 http://www.fernandomadalena.com/site_arquivos/918.pdf 

Alves D D, Paulino M F, Backes A A, Valadares Filho S C, Rennó L N 2004 Características de Carcaça de Bovinos Zebu e Cruzados Holandês-Zebu (F1) nas Fases de Recria e Terminação, Revista Brasileira de Zootecnia 33, n.5, p.1274-1284 

ASBIA 2011 Index ASBIA Importação, exportação e comercialização de sêmen http://www.asbia.org.br/novo/upload/mercado/relatorio2011.pd 

Azevedo M, Pires M F A, Saturnino H M, Lana A M Q,  Sampaio I B M, Monteiro J B N and Morato L E 2005 Estimativa de níveis críticos superiores do índice de temperatura e umidade para vacas leiteiras 1/2, 3/4 e 7/8 Holandês-zebu em lactação, Revista Brasileira de Zootecnia 34:2000-2008 http://www.scielo.br/pdf/rbz/v34n6/27254.pdf 

Balieiro Neto G, Ferreira J J and De Paz C C P 2007 Efeito do grupo genético de bovinos mestiços da raça Holandesa versus Gir sobre o ganho de peso de fêmeas em diferentes idades, Boletim de Indústria Animal 64:97-105 

Barbosa P F 2001 Sistemas mistos de produção de leite e carne bovina In Anais do Simpósio sabre Manejo, Nutrição e Sanidade de Gado Leiteiro São Carlos, SP p. 211-232 

Barbosa S B P, R P, Monardes H G, Dias F M, Santos D C and Batista A M V 2008 Milk and fat production of crossbred Holstein-Gir cows (Bos taurus taurus-Bos taurus indicus) in the Agreste region of the Brazilian state of Pernambuco, Genetics and Molecular Biology 31, 2, 468-474 http://www.scielo.br/pdf/gmb/v31n2/a12v31n2.pdf 

Boligon A A, Rorato P R N, Ferreira G B B, Weber T, Kippert C J and Andreazza J 2005 Herdabilidade e tendência genética para as produções de leite e de gordura em rebanhos da raça Holandesa no Estado do Rio Grande do Sul, Revista Brasileira de Zootecnia 34:1512-1518 http://www.revista.sbz.org.br/artigo/index.php?artigo=4273 

Caldas R P and Madalena  F E  2001 Ordenha com ou sem bezerro. In: Producão de Leite e Sociedade. Madalena F E, Matos L L and Holanda E V (Eds.), FEPMVZ, Belo Horizonte, pp. 243 – 260 http://www.fernandomadalena.com/site_arquivos/915.pdf 

Camilo B S, Freitas A F, Costa C N, Souza M A and Amaral TM 2009 Desempenhos produtivos e reprodutivos de vacas da raça Girolando, 46^a Reunião Anual da Sociedade Brasileira de Zootecnia, CD 

Cardoso V L, Nogueira J R and Arendonk J A M 1999 Optimum replacement and insemination policies for crossbred cattle (Holstein Friesian x Zebu) in the south-east region of Brazil, Livestock Production Science 58: 95–105 

Carvalho M P, Martins P C, Wright, J T C and Spers R G 2007 Cenários para o leite no Brasil em 2020, Embrapa Gado de Leite, Juiz de Fora, 190 p. 

Carvalho G R 2010 A Indústria de laticínios no Brasil: passado, presente e futuro. Circular Técnica 102, Embrapa Gado de Leite. http://www.cnpgl.embrapa.br/nova/livraria/abrir_pdf.php?id=26 

Carvalho G R and Carneiro A V 2010 Principais indicadores. Leite e derivados Vol 3 No. 30 Dec 2010 http://www.cileite.com.br/sites/default/files/2010_12_indicadores_leite_0.pdf 

Coelho M C 2004 Fazenda Santa Luzia Anais do 5⁰ Encontro de Produtores de Gado Leiteiro F1 pp.143-150 http://www.fernandomadalena.com/site_arquivos/867.pdf 

Combellas J 2006 Progress in methods for tropical cattle milking and calf rearing: implications for breeding objectives. 8th World Congress on Genetics Applied to Livestock Production, August 13-18, 2006, Belo Horizonte, MG, Brasil, CD, paper 31-06 

Costa C N, Blake R W, Pollak E J, Oltenacu P A, Quaas R L and Searle S R 2000 Genetic analysis of Holstein cattle populations in Brazil and the United States, Journal of Dairy Science 83:2963–2974 http://www.journalofdairyscience.org/article/S0022-0302(00)75196-4/abstract 

Costa C N 2005 Impacto das importações de sêmen na melhoria genética da raça Holandesa no Brasil, Milk Point, Aug 18 2005 http://www.milkpoint.com.br/radar-tecnico/melhoramento-genetico/impacto-das-importacoes-de-semen-na-melhoria-genetica-da-raca-holandesa-no-brasil-25031n.aspx 

Costa C N, Cobuci J A, Valloto A A and Martins M F 2011 Sumário Nacional de Touros da  Raça Holandesa – 2011, Documentos 149, Embrapa Gado de Leite, Juiz de Fora-MG http://www.cnpgl.embrapa.br/nova/informacoes/melhoramento/holandesa/doc149.pdf 

Dickerson G E 1969 Experimental approaches to utilizing breed resources, Animal Breeding Abstracts 37:191-202

EMBRAPA Gado de Leite 2011 http://www.cnpgl.embrapa.br/ 

EPAMIG 2010 Vacas F1 Holandês x Zebu: produção eficiente de leite, Informe Agropecuário, Belo Horizonte, v. 31 (358), 112 pp.  

Estado de Minas, newspaper Jul 11 2011 PIG completa uma ano com balanço positivo, In: Caderno Agropecuário, p. 4-5 

Facó O, Lobo R N B, Martins Filho R and Moura A A A 2002 Análise do desempenho produtivo de diversos grupos genéticos Holandês x Gir no Brasil, Revista Brasileira de Zootecnia 31:1944-1952 http://www.revista.sbz.org.br/artigo/index.php?artigo=3466 

Facó O, Lobo R N B, Martins Filho R, Oliveira S M. and Martins G A 2007 Heterogeneidade de (co)variância para a produção de leite nos grupos genéticos formadores da raça Girolando Revista Ciência Agronômica 38:304-309 http://www.ccarevista.ufc.br/seer/index.php/ccarevista/article/view/120 

Facó O, Lobo R N B, Martins Filho R, Martins G A, Oliveira S M. and Azevedo D M M R 2008 Efeitos genéticos aditivos e não-aditivos para características produtivas e reprodutivas em vacas mestiças Holandês × Gir, Revista Brasileira de Zootecnia 37:48-53 http://www.revista.sbz.org.br/artigo/index.php?artigo=6428 

Facó O, Martins Filho R, Lobo R N B, Azevedo D M M R and Oliveira S M P 2009 Efeito da redução da variação da duração de lactação na avaliação genética de bovinos leiteiros mestiços, Revista Ciência Agronômica 40:287-292 http://www.redalyc.org/src/inicio/ArtPdfRed.jsp?iCve=195318233017 

FAEG Federação da Agricultura e Pecuária de Goiás 2009 Diagnóstico da Cadeia Produtiva do Leite de Goiás: relatório de pesquisa. 64 p. 

FAEMG 2006 Diagnóstico da pecuária leiteira do Estado de Minas Gerais em 2005: relatório de pesquisa. – Belo Horizonte: 156 p. 

Falconer D S 1960 An Introduction to Quantitative Genetics. Oliver and Boyd, Edinburgh. 

FAOSTAT 2011 http://faostat.fao.org/site/569/DesktopDefault.aspx?PageID=569#ancor 

Faria F J C,  Vercesi Filho A E, Madalena F E and Josahkian L A 2009 Pedigree analysis in the Brazilian Zebu breeds, Journal of Animal Breeding and Genetics 126:148–153 

Ferreira J J and Madalena F E 1997 Efeito do sistema de cruzamento sobre o desempenho produtivo e reprodutivo de vacas leiteiras, Arquivos Brasileiros de Medicina Veterinária e Zootecnia 49:741-752 http://www.fernandomadalena.com/site_arquivos/040.pdf 

Ferreira J J and Ferreira M B D 1998 Sistema de produção de leite da EPAMIG – desempenho por grupo racial e custo de produção de leite,In: 2⁰ Encontro de produtores de gado leiteiro F₁, Cadernos Técnicos da Escola de Veterinária da UFMG 25:19-28 

Ferreira M B D, Lopes B C and Machado L H 2001 Sistema de produção de novilhas F1 com inseminação artificial, In: Produção de Leite e Sociedade, Madalena F E, Matos  L L and Holanda E V (Eds.),: FEPMVZ Editora, Belo Horizonte, pp. 362-378 http://www.fernandomadalena.com/site_arquivos/922.pdf 

Figueiredo P C, Freire N M S and Grisi L 1985 Variação da parasitose por Eimerias em bovinos Holando-zebu de acordo com a faixa de idade dos hospedeiros, Atas Sociedade de Biologia de Rio de Janeiro, 25:83-88 

Fiúza T L 2001 Produção de gado F1, In: Produção de Leite e Sociedade Madalena F E, Matos L L and Holanda E V (Eds.), FEPMVZ Editora, Belo Horizonte, pp. 379-381 http://www.fernandomadalena.com/site_arquivos/923.pdf 

Flores A A, Madalena F E and Teodoro R. L 2004 Desempenho comparativo de seis grupos de cruzamento Holandês/Guzerá. 12. Ganho de peso de bezerras e novilhas, Revista Brasileira de Zootecnia 33: 1695 - 1702 http://www.fernandomadalena.com/site_arquivos/015.pdf 

Freitas M S, Durães M C, Freitas, A F and Barra R B 2001 Comparação da produção de leite, gordura e duração da lactação em cinco graus de sangue originados de cruzamento entre Holandês e Gir em Minas Gerais,  Arquivos Brasileiros de Medicina Veterinária e Zootecnia 53:708-713 http://www.scielo.br/scielo.php?script=sci_arttext&pid=S0102-09352001000600017 

Freitas A, Freitas M, Teixeira N and  Costa C 2006  Additive genetic and heterosis effects on milk yield and fertility of Girolando cows  Proceedings of the 8th World Congress on Genetics Applied to Livestock Production, August 13-18, , Belo Horizonte, MG, Brasil, CD 

Gardner C O and Eberhart S A 1966  Analysis and interpretation of the variety cross diallel and related populations Biometrics 22:439-452

Goulart I L, Ferreira A M,  Ramos A A, Viana J H M, Sá W F, Camargo L S and Nogueira L A G 2009 Embriões F1 (Hol X Gir) produzidos in vitro utilizados a fresco em rebanhos leiteiros comerciais e controlado, Revista Brasileira de Ciência Veterinária 16: 77-82 http://www.uff.br/rbcv/site/index.php/numeros/view/16 

Grossi S F and Freitas M A R 2002 Eficiência reprodutiva e produtiva em rebanhos leiteiros comerciais monitoradospor sistema informatizado Revista Brasileira de Zootecnia 31:1362-1366 http://www.revista.sbz.org.br/artigo/index.php?artigo=3178 

Guimarães P H, Madalena F E and Cezar I M 2006 Comparative economics of Holstein/Gir F1 dairy female production and conventional beef cattle suckler herds - A simulation study, Agricultural Systems. 88:111-124 http://www.fernandomadalena.com/site_arquivos/001.pdf 

Herd  RM 1995 Effect of divergent selection for yearling growth rate on the maintenance feed requirements of mature Angus cows, Livestock Production Science 41:39-49  

Hickman C G 1979 The estimation and use of non-additive genetic variability in cattle and buffalo. Indian Journal of Animal Genetics and Breeding 1:1–6 

Honer M R, Braga R M and Rodrigues M L 1981 Analysis of the mean burdens of Dictyocaulus viviparus and Trichuris discolor in male calves, Holstein-Friesian and Holstein-Friesian x Guzerá, Pesquisa Veterinária Brasileira 1:133-135 

Hott M C, Souza R C S N P and Zoccal R 2009 Análise mesorregional da produção de leite no Brasil na última década, Panorama do leite online, Ano 3, no. 30. http://www.cileite.com.br/panorama/especial30.html 

Jorge A M, Fontes C A A and Paulino M S 1999 Composição corporal de bovinos de quatro raças zebuínas, abatidos em diferentes estádios de maturidade Revista Brasileira de Zootecnia 28:388-394 http://www.revista.sbz.org.br/artigo/index.php?artigo=2257 

Junqueira F S, Madalena F E and Reis G L 2005 Production and economic comparison of milking F1 Holstein x Gir cows with and without the stimulus of the calf. Livestock Production Science. 97:241 - 252 http://www.fernandomadalena.com/site_arquivos/703.pdf 

Junqueira F S, Madalena F E, Freitas L S, Valente B D, Calonge D and Silva B O 2010 Effects of introducing New Zealand Friesian genes into a Holstein/Gir dairy herd, Proceedings of the 9^th Congress on Genetics Applied to Livestock Production, WCGALP, Leipzig, 2010 CD 

Kahi A K, Kosgey I S, Cardoso V L and Van Arendonk J A M 1998 Influence of production circumstances and economic evaluation criteria on economic comparison of breeds and breed crosses,  Journal of  Dairy Science 81:2271–2279 http://www.journalofdairyscience.org/article/S0022-0302(98)75807-2/abstract 

Lemos A M, Teodoro R L, Barbosa R T, Freitas A F F and Madalena F E 1984  Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 1. Gestation length and birth weight, Animal Production 38:157 - 164 http://www.fernandomadalena.com/site_arquivos/031.pdf 

Lemos A M, Teodoro R L, Oliveira G P and Madalena F E 1985 Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 3. Burdens of Boophilus microplus under field conditions, Animal Production 41:187 - 191 http://www.fernandomadalena.com/site_arquivos/036.pdf 

Lemos A M, Madalena F E, Teodoro R L, Barbosa R T and Monteiro J B N 1992 Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 5. Age at first calving, Revista Brasileira de Genética, 15:73 - 83 http://www.fernandomadalena.com/site_arquivos/024.pdf 

Lemos A M, Teodoro R L and Madalena F E 1996 Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 9. Stayability, herd life and reasons for disposal, Revista Brasileira de Genética 19:259 - 264 http://www.fernandomadalena.com/site_arquivos/014.pdf 

Lemos A M, Verneque R S, Teodoro R L, Novaes L P, Gonçalves T M , Monteiro, J B N 1997 Efeito da estratégia de cruzamentos sobre características produtivas e reprodutivas em vacas do sistema mestiço do CNPGLEMBRAPA, Revista Brasileira de Zootecnia 26:704-708 http://www.revista.sbz.org.br/artigo/index.php?artigo=101 

Lôbo R B, Duarte F A M and Bezerra, L A F 1979a Phenotypic and genetic parameters of butterfat production in Pitangueiras cattle in Brazil, Revista Brasileira de Genética 2:161-170 http://web2.sbg.org.br/gmb/edicoesanteriores/v02n2/pdf/a06v02n2.pdf 

Lôbo R B, Duarte F A M and Bezerra, L A F 1979b Factors affecting milk production Pitangueiras cattle in Brazil, Revista Brasileira de Genética 2:199-209 http://web2.sbg.org.br/gmb/edicoesanteriores/v02n3/pdf/a03v02n3.pdf 

Lôbo R B, Reis J C, Duarte F A M and Wilcox C J 1988 Reproductive performance of Pitangueiras cattle in Brazil, Revista Brasileira de Genética 11:51-61 http://web2.sbg.org.br/gmb/edicoesanteriores/v11n1/pdf/a05v11n1.pdf 

Machado L H, Lopes B C and Ferreira M B D 2002  Sistema de produção de leite com matrizes F1 da Fazenda do Riacho, In: Anais do 4⁰ Encontro de Produtores de Gado Leiteiro F1, Belo Horizonte, 2002 pp.76-80  http://www.fernandomadalena.com/site_arquivos/850.pdf 

Madalena  F E, Freitas A F F and Martinez M L 1980 Evaluación comparativa de la producción de leche en vacas Holandesas y mestizas Holandés x Gir. Memorias de la 4a. Conferencia Mundial de Produccion Animal, Buenos Aires, 1980. v.2. p.650 - 658 http://www.fernandomadalena.com/site_arquivos/039.pdf 

Madalena F E 1981 Crossbreeding strategies for dairy cattle in Brazil, World Animal Review 38:23-30 http://www.fernandomadalena.com/site_arquivos/029.pdf 

Madalena F E, Valente J ,  Teodoro R L and Monteiro J B N 1983 Produção de leite e intervalo entre partos de vacas HPB e mesticas HPB:Gir num alto nível de manejo, Pesquisa Agropecuária Brasileira 18:195–200  

Madalena F E, Verneque R S and Teodoro R L 1985 Fatores que influenciam os preços do sêmen importado. Revista Brasileira de Genética, 8:377-384 http://www.fernandomadalena.com/site_arquivos/407.pdf 

Madalena F E 1988 A note on the effect of variation of lactation length on the efficiency of tropical cattle selection for milk yield. Theoretical and Applied Genetics 76:830 - 834 http://www.fernandomadalena.com/site_arquivos/202.pdf 

Madalena F E 1989 Cattle breed resource utilization for dairy production in Brazil, Revista Brasileira de Genética 12:183 – 220 http://www.fernandomadalena.com/site_arquivos/104.pdf 

Madalena F E, Teodoro R L, Nogueira J D, Moreira D P 1989a Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 4. Rate of milk flow, ease of milking and temperament, Revista Brasileira de Genética 12:39 - 51 http://www.fernandomadalena.com/site_arquivos/030.pdf 

Madalena F E, Lemos A M,  Teodoro R L, Monteiro J B N and Barbosa R T 1989b Causes terminating lactation records in Holstein-Friesian x Guzera crosses, Revista Brasileira de Genética 12:161- 167 http://www.fernandomadalena.com/site_arquivos/028.pdf 

Madalena F E, Lemos A M and Teodoro R L 1989c Effect of cost-price structure on the relative economic performance of alternative dairy cattle crossbreeding strategies, Revista Brasileira de Genética 12:887- 893 http://www.fernandomadalena.com/site_arquivos/027.pdf 

Madalena F E, Norte A L, Santos A J R, Freitas A F F 1989d Desenvolvimento do Mestiço Leiteiro Brasileiro (MLB). 2. Comparação do crescimento ponderal de tourinhos MLB e 1/4 Chianina x 3/4 Nelore. Arquivo Brasileiro de Medicina Veterinária e Zootecnia 41:355 - 367, 1989. http://www.fernandomadalena.com/site_arquivos/105.pdf 

Madalena F E, Lemos A M, Teodoro R L, Barbosa R T and Monteiro J B N 1990a Dairy production and reproduction in Holstein-Friesian x Guzera crosses, Journal of Dairy Science 73:1872 - 1886. http://www.journalofdairyscience.org/article/S0022-0302(90)78868-6/abstract 

Madalena F E, Teodoro R L, Lemos A M,  Monteiro J B N and Barbosa R T 1990b Evaluation of strategies for crossbreeding of dairy cattle in Brazil, Journal of Dairy Science 73:1887 - 1901 http://www.journalofdairyscience.org/article/S0022-0302(90)78869-8/abstract 

Madalena F E, Paiva J A J and Teodoro R L 1992a Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 6. Breed additive and heterosis effects on components of feed conversion efficiency in heifers, Revista Brasileira de Genética 15:595-601 http://www.fernandomadalena.com/site_arquivos/025.pdf 

Madalena F E, Lemos A M and Teodoro R L 1992b Consequences of removing the variation in lactation length on the evaluation of dairy cattle breeds and crosses. Revista Brasileira de Genética.15:585 - 594 http://www.fernandomadalena.com/site_arquivos/200.pdf 

Madalena F E 1993a La Utilizacion Sostenible de Hembras F1 en la Producción del Ganado Lechero Tropical Estudio FAO: Producción y Sanidad Animal (FAO), no. 111 / Roma (Italy), FAO, 103 p. http://www.fernandomadalena.com/livro_hembras_f1.html 

Madalena F E 1993b A simple scheme to utilize heterosis in dairy cattle, World Animal Review 74:17 – 25 http://www.fao.org/ag/AGa/AGAP/FRG/FEEDback/War/u9550b/u9550b0a.htm#TopOfPage 

Madalena F E, Teodoro R L, Lemos A M and Barbosa R T 1995 Comparative performance of six Holstein-Friesian x Guzerá grades in Brazil. 8. Calf mortality,  Revista Brasileira de Genética 8:215 - 220 http://www.fernandomadalena.com/site_arquivos/022.pdf 

Madalena F E, Abreu C P, Sampaio I B M and Ferreira Sobrinho F 1997a Práticas de cruzamentos em fazendas leiteiras afiliadas à Cooperativa Central de Produtores Rurais de Minas Gerais. Revista da  Sociedade Brasileira de Zootecnia. 26:924-934 http://www.fernandomadalena.com/site_arquivos/041.pdf 

Madalena F E, Madureira A P and Silvestre, J. R. A 1997b Características dos cruzamentos F1 para produção de leite em Minas Gerais. Cadernos Técnicos da Escola de Veterinária (UFMG). 18:41 - 52 http://www.fernandomadalena.com/site_arquivos/806.pdf 

Madalena F E 1999 Dairy cattle breeding program in Brazil. Development of the Brazilian Milking Hybrid In: FAO/ICAR Workshop on Breeding Strategies for Lower Input Animal Production Environments, 1999, Bella, Itália. P.365-378 http://www.fernandomadalena.com/site_arquivos/100.pdf 

Madalena F E 2001a A cadeia do leite no Brasil, In: Produção de Leite e Sociedade, Madalena F E, Matos L L and Holanda Jr. E V (eds) FEPMVZ Belo Horizonte, p. 1-26  http://www.fernandomadalena.com/indice_36.html 

Madalena F E 2001b Consideraciones sobre modelos para la predicción del desempeño de cruzamientos en bovinos. Archivos Latinoamericanos de Producción Animal. 9:108-117 

Madalena F E 2002 Bos indicus breeds and Bos indicus x Bos taurus crosses In: Encyclopedia of Dairy Sciences.1^st ed. : Academic Press., 2002, p. 576-585.http://www.fernandomadalena.com/site_arquivos/720.pdf 

Madalena F. E, Teodoro R L and Madureira A P 2003 Relationships of weight and height with age in hybrid Holstein-Friesian/Guzerá females. Genetics and Molecular Research. 2:271 - 278 http://www.fernandomadalena.com/site_arquivos/006.pdf  

Madalena F E and Junqueira F S 2004 The value of sexed bovine semen, Journal of Animal Breeding and Genetics 121:253 - 259 http://www.fernandomadalena.com/site_arquivos/704.pdf 

Madalena F E 2008 How sustainable are the breeding programs of the global main stream dairy breeds? - The Latin-American situation. Livestock Research for Rural Development. Volume 20, Article #19 http://www.lrrd.org/lrrd20/2/mada20019.htm

Madureira A P, Madalena F E and Teodoro R L 2002 Desempenho comparativo de seis grupos de cruzamento Holandês/Guzerá. 11. Peso e altura de vacas e novilhas. Revista Brasileira de Zootecnia. 31:658 - 667 http://www.fernandomadalena.com/site_arquivos/008.pdf 

Marcatti Neto A, Ruas J R M, Amaral R and Menezes A C 2004 Bezerros terminais de corte podem viabilizar sistemas de produção de leite, Informe Agropecuário, 25:25-31 

Mariante A S, McManus C and Mendonça J F 2003 Country report on the state of animal genetic resources. Brazil. http://dad.fao.org/en/Home.htm accessed March 2005 

Martinez M L,  Lee A J and Lin C Y 1988 Age and Zebu-Holstein additive and heterotic effects on lactation performance and reproduction in Brazil, Journal of Dairy Science 71: 800-808 http://www.journalofdairyscience.org/article/S0022-0302(88)79620-4/abstract 

Martins G A, Madalena F E, Bruschi J H, Costa J L, and Monteiro J B N 2003 Objetivos econômicos de seleção de bovinos de leite para fazenda demonstrativa na Zona da Mata de Minas Gerais, Revista Brasileira de Zootecnia 32:304-314 http://www.fernandomadalena.com/site_arquivos/403.pdf      Include URL

Martins G A, Madalena F E, Bruschi J H, Costa J L, Teodoro R L and Monteiro J B N 2004 Estimativas de parâmetros de cruzamentos para peso de fêmeas Holandês/Gir, Revista Brasileira de Zootecnia 33:1703-1710 http://www.fernandomadalena.com/site_arquivos/005.pdf 

Martins P C 2011 Quem desiste de produzir Leite? MilkPoint 28.11.2011 http://www.milkpoint.com.br/mypoint/paulomartins/p_quem_desiste_de_produzir_leite_3807.aspx 

MilkPoint 2011 ALB: consumo de lácteos cresce 60% em 30 anos. http://www.milkpoint.com.br/mercado/giro-lacteo/alb-consumo-de-lacteos-cresce-60-em-30-anos-69881n.aspx 

Moraes A C A, Coelho S G, Ruas J R M, Carvalho J C V, Vieira F A P and Menezes A C 2004. Estudo técnico e econômico de um sistema de produção de leite com gado mestiço F1 Holandês-Zebu, Arquivo Brasileiro de Medicina Veterinária e Zootecnia 56:745-749 http://www.scielo.br/pdf/abmvz/v56n6/a08v56n6.pdf 

Negrão J A 2008 Hormone release and behavior during suckling and milking in Gir, Gir × Holstein, and Holstein cows, Journal of Animal Science 86(Suppl. 1):21–26 http://jas.fass.org/cgi/content/full/86/13_suppl/21 

Nelsen T C, Short R E, Reynolds W L and Urick J J 1985 Palpated and visually assigned condition scores compared with weight, height and heart girth in Hereford and crossbred cows. Journal of Animal Science 60:363-368 http://jas.fass.org/content/60/2/363.full.pdf+html 

Novaes L P, Teodoro R L, Lemos A, M, Verneque R S and Monteiro J B N 1998 Desempenho produtivo e reprodutivo de animais de vários graus de sangue no sistema de produção da Embrapa-Gado de Leite. In: 2⁰ Encontro de produtores de gado leiteiro F₁, Cadernos Técnicos da Escola de Veterinária da UFMG 25:29-35 

NRC National Research Council 2000 Nutrient requirements of beef cattle, 7^th Revision Update www.nap.edu 

Oliveira  G P and Alencar M M 1990 Resistência de bovinos de seis graus de sangue Holandês-Guzerá ao carrapato (Boophilus microplus) e ao berne (Dermatobia hominis), Arquivo Brasileiro de Medicina Veterinária e Zootecnia 42:127-135 

Paiva J A J, Madalena F E, Teodoro R L and Campos A T 1992 Food conversion efficiency in six groups of Holstein-Friesian/Zebu crosses, Livestock Production Science 30:213 - 222 http://www.fernandomadalena.com/site_arquivos/026.pdf 

Paloschi C G 1981  Carga média, relações fêmeas/machos e número de ovos por fêmea, em populações de Cooperia sp, em bezerro Holandês e mestiços de Holandês x Guzerá, M. Sc. Thesis, Universidade Federal Rural de Rio de Janeiro, 91 p.   

Peixoto M G C D, Verneque R S, Teodoro R L,  Penna V M and Martinez M L 2006 Genetic trend for milk yield in Guzerat herds participating in progeny testing and MOET nucleus schemes, Genetics and Molecular Research 5:454-465 http://www.geneticsmr.com//year2006/vol5-3/pdf/gmr0215.pdf  

Peixoto M G C D, Panetto J C C, Bruneli F A T, Pires M F A,  Campos B C, Lopes F C F, Zoccal R, Hott M C and Verneque R S 2011 Brazilian dairy cattle and production – an overview, Unpublished  

Peixoto M G C D,  Poggian C F, Verneque R S, Egito A A, Carvalho M R S, Penna V M, Bergmann J A G, Viccini L F and Machado M A 2012 Genetic Basis and inbreeding in the Brazilian Guzerat (Bos indicus) subpopulation selected for milk production. Livestock Production Science 131:168-174 

Penna V M, Verneque R S, Teodoro R L,  Melo V J M and Madalena F E 2002 Utilization of a MOET nucleus in the improvement of the Guzerá. Proceedings of the 7th World congress on Genetics Applied to Livestock Production. Montpellier: WCGALP, CD, p.25-14 http://www.fernandomadalena.com/site_arquivos/706.pdf 

Penna V M, Melo V J M, Teodoro R L, Verneque R S and Peixoto M G C D 2005 Situação atual e potencialidades da raça Guzerá na pecuária leiteira nacional In: Aspectos econômicos, técnicos, sociais e ambientais da atividade leiteira. Juiz de Fora: Embrapa Gado de Leite, 2005. p.103-110  

Pereira M E G, Ruas J R M, Queiroz D S, Silva E A, Soares Jr J A G and Silva G W V 2010 Avaliação da produção de bezerros de corte em um rebanho leiteiro F1 Holandês x Zebu, Anais da 47^a Reunião Anual da Sociedade Brasileira de Zootecnia, Salvador, BA, CD 

Peroto D, Castanho M J P, Rocha J L and Pinto J M 1997 Descrição das curvas de crescimento de fêmeas bovinas Guzerá, Gir, Holandês x Guzerá e Holandês x Gir, Revista Brasileira de Zootecnia 26:283-288 http://www.revista.sbz.org.br/artigo/index.php?artigo=40 

Perotto D, Kroetz I A and Rocha J L 2010 Milk production of crossbred Holstein × Zebu cows in the northeastern region of Paraná State, Revista Brasileira de Zootecnia 39:758-764 http://www.revista.sbz.org.br/artigo/index.php?artigo=8533 

Pinha L C, Carvalho G R and Travassos G F  2010 Poder de compra e consumo de lácteos no Brasil, 8⁰ Congresso internacional do leite, Juiz de Fora http://www.cileite.com.br/sites/default/files/2010_07_poder%20de%20compra%20e%20consumo.pdf 

Pontes J H F, Silva K C F, Basso A C, Rigo A G, Ferreira C R, Santos G M G, Sanches B V, Porcionato J P F, Vieira P H S, Faifer F S, Sterza F A M, Schenk J L and Seneda M M 2010 Large-scale in vitro embryo production and pregnancy rates from Bos taurus, Bos indicus, and indicus-taurus dairy cows using sexed sperm, Theriogenology 74:1349–1355 

Porcionato M A F, Negrão J A and Lima M L P 2005 Produção de leite, leite residual e concentração hormonal de vacas Gir × Holandesa e Holandesa em ordenha mecanizada exclusiva, Arquivo Brasileiro de Medicina Veterinária e Zootecnia, 57:820-824 http://www.scielo.br/scielo.php?pid=S0102-09352009000200001&script=sci_arttext  

Preston T R and Vaccaro L 1989  Dual purpose cattle production systems. In: Phillips, C.J.C. (Ed.), New Techniques in Cattle Production. Butterworths Scientific, London, pp. 20 – 32 

Razook A G, Leme P R, Packer I U, Luchiari Filho A, Inardon R F, Trovo J B F, Capelozza C N Z, Pires F L, Inascimento J, Barbosa C, Coutinho G L B, Oliveira I W J 1986. Evaluation of Nelore, Canchim, Santa Gertrudis, Holstein, Brown Swiss and Caracu as sire breeds in matings with Nelore cows: effects on progeny growth, carcass traits and crossbred productivity. In: Proceedings of the 3^rd World Congress on Genetics Applied to Livestock Production, Lincoln, NE, University of Nebraska, v. 9, p. 348-352. 

Regitano L C A, Martinez M L and Machado M A 2006 Molecular aspects of bovine tropical adaptation, 8th World Congress on Genetics Applied to Livestock Production, August 13-18, Belo Horizonte, MG, Brasil, CD 

Rhoad A O 1935 Production of the Brazilian dairy cattle under the penkeeping system, Zeitschrift fur Tierzuchtung und Zuchtungsbiologie  33:105-118 

Reis Filho J C, Lopes P S, Verneque R S, Torres R A, Teodoro R L and Carneiro P L S 2010 Population structure of Brazilian Gyr dairy cattle, Revista Brasileira de Zootecnia 39:2640-2645 http://www.revista.sbz.org.br/artigo/visualizar.php?artigo=66014 

Rotz C A, Soder K J, Skinner R H, Dell C J, Kleinman P J, Schmidt J P and Bryant R B 2009 Grazing can reduce the environmental impact of dairy production systems, Online. Forage and Grazinglands doi:10.1094/FG-2009-0916-01-RS. 

Ruas J R M, Silva M A, Ferreira J J, Amaral R, Campos B C, Menezes A C and Chagas G F 2008 Desempenho produtivo e reprodutivo de vacas F1 Holandês x Zebu em rebanhos da EPAMIG, In 6⁰ Encontro de Produtores de Gado Leiteiro F₁ PUC-Minas, Belo Horizonte 2008 pp 144-183 

Ruas J R M, Carvalho B C, Queiroz D S, Silva E A, Conceição P R M and Oliveira P A 2011 Influência do peso ao parto e do comportamento na produção de leite de vacas F1 Holandês x Zebu (F1 HZ) Anais da 48a Reunião Anual da Sociedade Brasileira de Zootecnia, Belém, PA, CD 

Rutledge J J 2001 Greek temples, tropical kine and recombination load, Livestock Production Science 68:171-179 

Santiago R L 2004 Projeto NELOGIR. Cruzamento Nelore x Gir como matriz para F1. Resultados parciais.  In: Anais do 5⁰ Encontro de Produtores de Gado leiteiro F1 FEPMVZ Editora, Belo Horizonte pp. 151-162 http://www.fernandomadalena.com/site_arquivos/868.pdf 

Santiago R L 2008 Produção de matrizes F₁ pela Fazenda Calciolândia a partir de matrizes ½ sangue Nelore-Gir Leiteiro, In: Anais do 6⁰ Encontro de Produtores de Gado leiteiro F1 PUC Minas, Belo Horizonte pp. 193-207 

Santos E P B, Montenegro A R, Neves R D, Silva R R, Campello C C and Oliveira S M P 2011 Avaliação de características produtivas de grupos genéticos Holandês x Gir de um rebanho bovino leiteiro no Município de Beberibe 48 Reunião Anual da Sociedade Brasileira de Zootecnia, CD 

Silva F F, Valadares Filho S V, Ítavo L C V, Veloso C M, Valadares R F D, Cecon P R, Paulino P V R and Moraes E H V K 2002  Exigências líquidas e dietéticas de energia, proteína e macroelementos minerais de bovinos de corte no Brasil, Revista Brasileira de Zootecnia 31:776-792http://www.revista.sbz.org.br/artigo/index.php?artigo=3314 

Silva M B G B, Paiva L C, Cembranelli M A R, Martins M F and Rodrigues W B R 2011 Programa de Melhoramento Genético da Raça Girolando - Sumário de Touros - Resultado do Teste de Progênie Junho/2011, Documentos 148, EMBRAPA Gado de Leite http://www.girolando.com.br/site/progenie/2011/Sumario-de-Girolando2011.pdf 

Soares M P, Gaya L G, Lorentz L H, Batistel F, Rovadoscki G A, Ticiani E, Zabot V,  Domenico Q, Madureira A P and Pértile S F N 2011 Relationship between the magnitude of the inbreeding coefficient and milk traits in Holstein and Jersey dairy bull semen used in Brazil, Genetics and Molecular Research 10:1942-1947http://www.geneticsmr.com//year2011/vol10-3/pdf/gmr1010.pdf                  

Solís J C, Byers F M, Schelling G T, Long C R and Greene L W 1988  Maintenance requirements and energetic efficiency of cows of different breeds, Journal of Animal Science 66:764-773http://jas.fass.org/search?submit=yes&pubdate_year=1988&volume=66&firstpage=764&doi=&author1=&author2=&title=&andorexacttitle=and&titleabstract=&andorexacttitleabs=and&fulltext=&andorexactfulltext=and&fmonth=&fyear=&tmonth=&tyear=

&format=standard&hits=10&sortspec=relevance&submit=yes&submit=Subm 

Souza G N, Brito, M A V P, Lange C C, Faria C G, Moraes L C D and Brito J R F 2009 Qualidade do leite de rebanhos bovinos localizados na Região Sudeste: Espírito Santo, Minas Gerais, Rio de Janeiro, janeiro/2007 a junho/2008 www.terraviva.com.br/clique/IIIcbql.pdf 

Teodoro R L, Lemos A M, Barbosa R T and Madalena F E 1984 Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 2. Traits related to the onset of the sexual function, Animal Production 38:165 - 170 http://www.fernandomadalena.com/site_arquivos/032.pdf 

Teodoro R L Lemos A M and Madalena F E 1994 Carga parasitaria de Boophilus microplus em vacas mestiças Europeu x Zebu, Revista Brasileira de Zootecnia 23:223-228 http://www.revista.sbz.org.br/artigo/index.php?artigo=443 

Teodoro R L, Madalena F E, Lemos A M and Anjos D A 1996a Comparative performance of six Holstein-Friesian x Guzera grades in Brazil. 10. Disposal value Revista Brasileira de Genética 9:417-420 http://www.fernandomadalena.com/site_arquivos/021.pdf 

Teodoro R L, Madalena F E and Smith C 1996b The value of F₁ dairy Bos taurus-Bos indicus embryos for milk production in poor environments, Journal of Animal Breeding and Genetics 113:471–482 

Teodoro R L, Lemos A M and Madalena F E 1998 Effect of ticks (Boophilus microplus) infestations on milk yield of Bos taurus x Bos indicus crosses. Proceedings 6th World Congrees on Genetics Applied to Livestock Production, Armidale, 27:177 - 180 http://www.fernandomadalena.com/site_arquivos/500.pdf 

Teodoro R L and Madalena F E 2002a Evaluation of crosses of Holstein, Jersey or Brown Swiss sires x Holstein-Friesian/Gir dams. 1. Dairy production and reproduction, Tropical Animal Health and Production 35:105-115. 

Teodoro R L and Madalena F E 2002b Evaluation of crosses of Holstein, Jersey or Brown Swiss sires x Holstein-Friesian/Gir dams. 2. Female liveweights, Genetics and Molecular Research 1:25-31 http://www.fernandomadalena.com/site_arquivos/009.pdf 

Teodoro R L and Madalena F E 2005 Evaluation of crosses of Holstein, Jersey or Brown Swiss sires x Holstein-Friesian/Gir dams. 3. Lifetime performance and economic evaluation, Genetics and Molecular Research 4:84-93 http://www.fernandomadalena.com/site_arquivos/002.pdf 

Vencovsky R, Dias O J and Ricardo Y 1970 Um modelo genético aplicado à produção de leite em gado bovino. In Relatório do Departamento de Genética, Escola Superior de Agricultura Luiz de Queiroz, Piracicaba SP, p. 130-136

Vercesi Filho A E, Madalena F E and Ferreira J J  2000 Pesos econômicos para seleção de gado de leite. Revista Brasileira de Zootecnia, v.29, p.145-152 http://www.fernandomadalena.com/site_arquivos/404.pdf 

Vercesi Filho A E, Faro L, Paz C C P, Cardoso V L, Silva D A L and Junqueira F S 2007 Fatores que influenciam o preço do sêmen de touros da raça Gir Leiteiro, Anais da 44a Reunião Anual da Sociedade Brasileira de Zootecnia, Jaboticabal,CD 

Vercesi Filho, A.E, Verneque R S, Peixoto M G C D, Machado M A, Penna V M and Cardoso V L 2010 Selection of tropical dairy cattle – The experience from the Brazilian Gyr and Guzerat,  Proceedings of the 9^th Congress on Genetics Applied to Livestock Production, WCGALP, Leipzig, 2010 CD 

Vercesi Filho A E, Balancin Jr A,  Zadra L F and Cardoso V L 2011 Desempenho produtivo e reprodutivo de vacas mestiças Holandês x Gir Leiteiro de diferentes grupos genéticos Anais da 48a Reunião Anual da Sociedade Brasileira de Zootecnia, Belém,CD

Verneque R S,  Peixoto M G C D, Machado M A, Teodoro R L and Werneck G A  2008 Melhoramento genético de gado de leite, In: Anais do VII Simpósio Brasileiro de Melhoramento Animal, São Carlos, SP http://sbmaonline.org.br/anais/vii/palestras/pdfs/palestra10.pdf

Viana J H M, Siqueira L G B, Palhão M P and Camargo L S A 2010 Use of in vitro fertilization technique in the last decade and its effect on Brazilian embryo industry and animal production, Acta Scientiae Veterinariae 38:661-674 http://www.ufrgs.br/actavet/38-suple-2/23_SBTE_JHENRIQUE.pdf 

Villares J B 1941 Climatologia Zootécnica. III. Contribuição ao estudo da resistência e susceptibilidade genética dos bovinos ao Boophilus microplus. Boletim da Indústria Animal 4:60-86  

World Bank 2011 http://www.google.com.br/publicdata/explore?ds=d5bncppjof8f9_&met_y=sp_pop_totl&idim=country:BRA&dl=pt-BR&hl=pt-BR&q=brasil+popula%C3%A7%C3%A3o 

Zoccal R and Carneiro A V  2008 Uma análise conjuntural da produção de leite brasileira, Panorama do leite online, Ano 2, no. 19 http://www.cnpgl.embrapa.br/panorama/conjuntura19.html 

Zoccal R 2011 Quantos são os produtores de leite no Brasil?, Balde Branco, Ano 48, No. 566, p. 36-38 

Appendix-Background on the Brazilian dairy industry

Milk prices were de-regulated in 1991, and since then international dairy processor firms established themselves in the Country and a process of fusion of both international and national firms and cooperatives has occurred. However many small dairy plants still operate (in 2005 there were 1973 of them). On-farm milk cooling and truck bulk transport has since been generalized, except in the less developed regions. Most milk (34%) is processed into cheese, 25% is sold fresh, 19% dried, 8% as other milk products and 14% is consumed on farms. About one third of the milk produced is not subjected to Federal inspection, although part of it may be inspected by State or municipality authorities, sold informally as milk, cheese and other products or self consumed. Since de-regulation milk processed by UHT and packed in Tetrapack boxes prevailed over the previous pasteurization and packing in plastic suchets. Currently about 75% of the inspected milk is UHT processed and 25% is pasteurized (Carvalho et al 2007). The switch to UHT milk caused the supermarkets to become the main milk distributors instead of the bakeries where pasteurised milk was previously sold. Concentration in the food distribution sector also occurred. The 10 largest supermarket chains were responsible for 70% of the food sales in 2009 against 49% in 1997. This oligopoly concentration gives the supermarkets great bargaining power on milk price determination (Carvalho  2010).

The 2006 census by the Brazilian Institute of Geography and statistics (IBGE) indicated nearly 5.2 million farms, a quarter of which (1.35 million) produced milk. However, there is a strong heterogeneity of dairy farm sizes and types. As it may be seen in Table A1, while 45% of the farms contributed with 4.6% of the total milk produced, 0.7% contributed with 16%. In the lower production class of <10 l/d most farms (68.7%) did not sell milk at all. Herd size in this class was 3 milked cows while it was 107 cows in the largest production class of >500 l/day (Zoccal, 2011).  Martins (2011) indicated that between 1996 and 2006 the number of farmers in the <50 l/day class was reduced by half a million and the number in the >200 l/d class by 20 thousand, while the number in the 50 to 200 l/d class increased by 60 thousand and was responsible for the increase in total national production. A trend to increase production per farm is favoured by the larger dairy processing plants via milk price bonus on quantity. Between 1999 and 2009, the ten largest plants increased the quantity of milk they processed from 34 to 42% of the annual total country milk and reduced the number of supplier farms from 128 to 76 thousand, but with higher average milk production, respectively 50 and 248 kg/d (Carvalho  2010). However, the farm-gate prices of milk were reduced (e.g. at a rate of 1.9% p.a. between 1996 to 2005, FAEMG 2006).

Climate

A climate map of Brazil is in Figure A1 and a rainfall map with isotherms in Figure A2. In central Brazil, where most milk production takes place, climate is characterised by humid hot summers and winter dry seasons, while in the South cold winters limit pasture production.

Regional distribution of production

Dairy production occurs mainly in the Southeast and South regions where most of the population lives (Figure A3). Minas Gerais is the main producer State, followed by Paraná and Rio Grande do Sul.  The less developed Northeast region ranks second in number of cows but milk yield per cow is low, as in the North (Table A2). As in other parts of the world, ruminant production was and continues to be displaced by agriculture to regions of cheaper land and labour and dairy production in Brazil has moved over the years from the more populated and developed regions to the West and to the North. Even within the Southeast and South regions a movement towards the West has occurred (Figure A4). The Southeast region reduced its relative participation from 55% of the total milk produced in 1975 to 35% in 2010, particularly in São Paulo State, which even had an absolute decline in production. In the decade 1996 to 2006, milk production increased at annual rates of 9.9, 5.1 and 3.9 % in the North, South and Center-West regions and 2.7 and 1.3 in the Northeast and Southeast. The growth of UHT milk and cheese consumption, which allow production away from the main consumer centres may be another cause for the regional changes as also agrarian reform settlements (Carvalho et al 2007) .

Milk quality

Based on some 190,000 samples, the geometric means of milk protein and fat content were 3.27 and 3.71% at EMBRAPA’s laboratory for farms in Minas Gerais, Rio de Janeiro and Espirito Santo States. Mean somatic cell counts (CCS) were 343 x 10³ cell/ml and total bacterial count means were 648 x 10³ colony forming units/ml (Carvalho 2010). About half the milk samples had CCS below 400 x10³ cells/ml and 80% below 750 x 10³ while 6% were below 100 x10³ colony forming units/ml and 42% below 750 x 10³.  Milk quality has not improved in the past recent years, in spite of a regulation by the Ministry of Agriculture establishing limits on the four items mentioned (Souza et al 2009). Currently only a few processor plants, including some of the largest, have attractive bonus payments on quality but many processors do not pay on quality at all.

Received 4 April 2012; Accepted 5 May 2012; Published 1 June 2012

Go to top