• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
PhD Thesis WM Rauw (2001)
 

PhD Thesis WM Rauw (2001)

on

  • 855 views

Changes in food resource allocation patterns with selection for litter size - a mouse model

Changes in food resource allocation patterns with selection for litter size - a mouse model

Statistics

Views

Total Views
855
Views on SlideShare
700
Embed Views
155

Actions

Likes
0
Downloads
1
Comments
0

7 Embeds 155

http://www.behance.net 78
http://wendyrauw.com 52
http://www.wendyrauw.com 10
http://www.linkedin.com 9
http://www.docshut.com 2
http://prosite.com 2
http://www.slashdocs.com 2
More...

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    PhD Thesis WM Rauw (2001) PhD Thesis WM Rauw (2001) Document Transcript

    • Changes in food resource allocation patterns with selection for litter size - a mouse model -Wendy Mercedes Rauw
    • Changes in food resource allocation patterns with selection for litter size - a mouse model - Wendy Mercedes RauwDoctor Agriculturae Thesis 2001Department of Animal ScienceAgricultural University of NorwayP.O. Box 5025, N-1432 ÅsISBN 82-7479-011-1
    • One of the most remarkable featuresin our domesticated racesis that we see in them adaptation,not indeed to the animal’s (…) own good,but to man’s use or fancy.Charles Darwin, 1872To my little Lenaand those who are yet to come
    • PrefaceThe work of this thesis has been performed between March 1997 and January 2001 in Ås, at theAgricultural University of Norway, Department of Animal Science and was financed by a grantfrom the Norwegian Research Council. Although this thesis has been mainly a personalenterprise, I would like to thank greatly some people for their help and stimulation. This work was initiated by Ella Luiting, whom I visited in Norway from May 1995 to January1996. Ella was the supervisor of my MSc thesis (Wageningen Institute of Animal Science) andshe stimulated and helped me in applying for the PhD-grant. During the course of my PhD-studyshe inspired me greatly. I am very thankful for her contribution and help along the years. Furthermore, I am most grateful to Pieter Knap and Martin Verstegen for their great helpand everlasting interest. Ella, Pieter and Martin have been my main supervisors and without themthis work would not have been as it is. They have contributed to the experiments and paperswith critical comments and suggestions and kept on stimulating me through the process. Thankyou all for your great friendship! I have visited Rolf Beilharz at the Department of Animal Production, Institute of Land andFood Resources in Melbourne, Australia during the European winter months of ´97/´98, todiscuss the Resource Allocation Theory. This trip has been very inspiring to me and I thank Rolfand his wife Vyrna for their hospitality and friendship and for showing me around their beautifulcountry. The first paper of this thesis, my very first article, was written in Wageningen under theguidance of Egbert Kanis, Elsbeth Noordhuizen-Stassen and Jan Grommers. I have enjoyed themonthly meetings a lot and to accomplish this paper has given me the confidence to keep onwriting. Thanks! The months of March and April 2000 I have spent at the Instituto Nacional deInvestigación y Tecnología Agraria y Alimentaria (INIA), Departamento Mejora Genética yBiotecnología in Madrid, Spain. At this institute I have written two of the chapters of thisthesis and I thank the institute for providing me the resources. Special thanks go to Mª CarmenRodriguez Valdovinos and to all the scientists and researchers at the INIA for their hospitality! The final work of my thesis was done at the Institut de Recerca i TecnologiaAgroalimentàries (IRTA), Àrea de Producció Animal in Lleida, Spain, where I work sinceSeptember 2000. Special thanks to José Luis Noguera Jiménez, Luis Varona Aguado, Nuria AlosSaiz and Meritxell Arqué Clemens. I have enjoyed my life in Norway a lot and I thank the personnel at the Department ofAnimal Science for being my colleagues, office mates and friends. Although I have conducted allexperiments myself, Kari Kjus who worked in the Department´s mouse laboratory has helpedwhenever needed. Tusen takk Kari for din hjelp! Then at last, but most of all, I want to thank my family in The Netherlands and in Spain fortheir interest in me and in my work. Nothing could be more stimulating and nothing is moreimportant to me. Luis Gomez Raya, my lover and best friend, I thank you for your love and forhaving with me the most beautiful daughter on earth. Lena was born in Norway in August 1999.As much pleasure I got from working on my thesis, nothing is comparable to the pleasure ofbeing her mother. Wendy Rauw, Lleida, December 2000
    • ContentsChapter 1 Introduction 1Chapter 2 Undesirable side effects of selection for high 5 production efficiencyChapter 3 Food resource allocation patterns in non- 35 reproductive males and females. I. Trends in body weight and food intake against timeChapter 4 Food resource allocation patterns in non- 49 reproductive males and females. II. Trends in body weight against food intakeChapter 5 Body composition in non-reproductive adult 63 males and femalesChapter 6 Behavioural strategies in non-reproductive 73 adult femalesChapter 7 Food resource allocation patterns in lactating 89 femalesChapter 8 General discussion 113Chapter 9 Recommendations for further research 133Summary 139
    • 1 INTRODUCTIONQuantitative genetics developed early in the last century with the rediscovery ofMendelian genetics (Lynch and Walsh, 1998). The major goal of applied quantitativegenetics is the identification of ‘elite’ individuals which serve as parents for the nextgeneration. Because parents pass on their genes and not their genotypes, an animal isassigned a ‘breeding value’, which refers to its genotypic value (Falconer and Mackay,1996). The current methodologies that are used to estimate these breeding valuesassume additivity and linearity of relationships between traits. However, thephenotypic relationship between traits is often found to be curvilinear. Sölkner andJames (1994) suggest that this may result from various physiological limitations andfeedback mechanisms, such as the situation where traits are competing for a limitedamount of resources. Resources come from food intake or body stores. Weiner (1992) proposed ‘the barrel model’ of an organism’s resource allocation pattern (Figure 1). Inputs constraints (food intake, digestion and absorption) are engaged in series whereas outputs (maintenance, growth, production) are parallel. If the sum of output rates does not match the input, the balance is buffered by energyFigure 1 The barrel model of an organism’s storage of the body. In the long run,energy balance (after Weiner, 1992). The however, energy expenditure mustfirst spigot always leaks (basal metabolic balance energy intake. According torate). FI = food intake; D = digestion; A = Beilharz et al. (1993), in a limitedabsorption; M = maintenance; G = growth; P = resource situation, when too manyproduction. resources are allocated towards one trait, less resources will be left forother resource demanding processes to perform optimally. This must result in anegative relationship between the traits. With artificial selection, when geneticallyforced to produce highly, resources may be reallocated from other processes, leavingthe animal lacking in ability to respond to other demands (Dunnington, 1990). As a consequence, genetic selection for increased levels of production is oftencompromised by unfavourable relationships between production and fitness traits andby limits to selection. This unfavourable relationship is aggravated when selection is onhigh production combined with high food efficiency, i.e., high production efficiency. Anexample of such an unfavourable relationship is a prolonged period of anoestrus afterweaning in genetically lean lines of pigs and an increased pre-weaning mortality rate of
    • 2 Introthe piglets. These problems result from excessive mobilisation of body reserves of thedam and problems of development and adaptation of their offspring (Ten Napel, 1996). To understand why selection for production traits is compromised by undesirableside effects and selection limits, research needs to be directed to the biologicalaspects of selection, or in other words, to the physiological processes on which a geneand thus genetic selection acts. Without this knowledge genetic improvement throughartificial selection is essentially a black box technique. Knowledge of biologicalbackgrounds will offer the opportunity to understand, anticipate and eventually avoidnegative side effects of selection. The present study investigates changes in food resource allocation patterns withlong-term selection for litter size using a mouse model. It is hypothesised that animalsselected for high litter size allocate a higher amount of resources to the trait selectedfor, leaving less resources to respond to other demands. The mice used in this studyoriginate from two lines of the Norwegian mouse selection experiment (Vangen, 1993): aline selected for more than 90 generations for high litter size at birth (S-line) and anunselected control line (C-line). The average number of live-born pups is about 20 in theS-line and 10 in the C-line.The subgoals of this study are: To investigate changes in resource allocation patterns in non-reproductive males and females. To investigate changes in body composition in relation to resource allocation patterns in non-reproductive males and females. To investigate changes in behavioural strategies in relation to resource allocation patterns in non-reproductive females. To investigate changes in resource allocation patterns in lactating females. To investigate offspring development in relation to resource allocation patterns.The outline of this study is as follows: Chapter 2 presents a general overview of undesirable side effects of selection forhigh production efficiency in broilers, pigs and dairy cattle, with respect to metabolic,reproduction and health traits. These side effects are discussed in relation to theresource allocation theory proposed by Beilharz et al. (1993). Chapters 3 to 6 focus on non-reproductive animals. In Chapter 3, changes in foodresource allocation patterns with selection for litter size are investigated in non-reproductive male and female mice of the C- and the S-line. Growth and food intakecurves are fitted to individual data. Differences in these curves between species orlines can mostly be explained by differences in mature size. Therefore, parameters arescaled by individual estimates of mature body weight. Differences that remain afterscaling are a consequence of what have been called ‘specific genetic factors’ (Taylor,1985). The presence of such specific genetic factors indicates if selection forincreased litter size has disproportionally changed the resource allocation pattern.
    • 1. Introduction 3 The existence of specific genetic factors for body weight and food intake isindicated by variation in efficiency parameters such as growth efficiency andmaintenance requirements. In Chapter 4, estimation of residual food intake (RFI)(Luiting and Urff, 1991) and Parks’ (1982) estimates of growth efficiency andmaintenance requirements are used to quantify these factors in the same individuals aswhich are used in Chapter 3. It can be suggested that the foetus acts as a parasite that would deplete the damfrom her energy stores (Hammond, 1944). Usually, during the first half of pregnancy,body protein and lipid content is increased for support of fetal growth and lactation(Luz and Griggio, 1996). Since S-line females have to support a litter size that has beenhighly increased by artificial selection, in Chapter 5 it is investigated whether bodycomposition at maturity has been affected as a correlated effect of selection for highlitter size, to sufficiently support the fetuses during pregnancy and lactation.Furthermore, part of the observed differences between individuals in RFI may beattributable to differing proportions of body protein and lipid (Luiting, 1990).Therefore, in Chapter 5, it is investigated if the differences between the C- and S-lines in body composition of non-reproducing adult males and females are consistentwith expectations from the observed differences between the lines in RFI (Chapter 4). Several studies have indicated that a higher RFI is related to a higher level ofactivity. Differences in activity may suggest underlying differences in coping behaviourin response to unexpected stresses. Chapter 6 investigates whether coping strategiesin non-reproductive females have been affected as a correlated effect of selection forincreased litter size. For this reason, females of the C- and the S-line are subjected tothree non-social tests and a social confrontation test. Chapter 7 focuses on reproductive females. Changes in food resource allocationpatterns with selection for litter size are investigated in lactating female mice of theC- and the S-line and related to offspring development. To manipulate experimentallythe energy burden of lactation, in each line, half of the females support litters that arestandardised at birth to eight pups per litter (when larger than eight pups), and half ofthe females support the natural litter size. Food resource allocation patterns arequantified by RFI. The consequences of food resource allocation patterns for offspringdevelopment was indicated by pre-weaning mortality rates and degree of maturity ofthe pups from birth to weaning. Chapter 8 is a general discussion of the results of Chapters 3 to 5 and Chapter 7.The introduction to this chapter describes how selection for high production efficiencymay compromise fitness. Thereafter, the main results of the four chapters are brieflysummarised and the results are discussed in relation to pig production. During thecourse of this study, body composition was measured in, and behavioural test werecarried out on lactating females of the C- and the S-line, but the results of these trialsare not included as single chapters in this thesis. However, the main results of the bodycomposition measurements on lactating females are included in Chapter 8. Chapter 9 concludes this study with recommendations for further research.
    • 4 IntroReferences• Beilharz, R.G., Luxford, B.G., Wilkinson, J.L., 1993. Quantitative genetics and evolution: Is our understanding of genetics sufficient to explain evolution? J. Anim. Br. Gen. 110:161-170.• Falconer, D.S. and Mackay, T.F.C., 1996. Introduction to quantitative genetics. 4th edition, Longman Group Ltd, UK.• Hammond, J., 1944. Physiological factors affecting birth weight. Proc. Nutr. Soc. 2:8-11.• Luiting, P., 1990. Genetic variation of energy partitioning in laying hens: causes of variation in residual feed consumption. World’s Poultry Sci. J. 46:133-152.• Luiting, P. and Urff, E.M., 1991. Optimization of a model to estimate residual feed consumption in the laying hen. Livest. Prod. Sci. 27:321-338.• Luz, J. and Griggio, M.A., 1996. Distribution of energy between food-restricted dams and offspring. Ann. Nutr. Metab. 40:165-174.• Lynch, M. and Walsh, B., 1998. Genetics and analysis of quantitative traits. Sinauer Associates.• Parks, J.R., 1982. A theory of feeding and growth of animals. Springer-Verlag, Berlin.• Sölkner, J. and James, J.W., 1994. Curvilinearity in the relationship between traits competing for resources: a genetic model. Proc. of the 5th World Congr. Gen. Appl. Livest. Prod., Guelph, 19:151-154.• Taylor, St C.S., 1985. Use of genetic size-scaling in evaluation of animal growth. J. Anim. Sci. 61: (Suppl. 2) 118-141.• Ten Napel, J., 1996. Genetic aspects of intervals from weaning to estrus in swine. PhD thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Vangen, O., 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.• Weiner, J., 1992. Physiological limits to sustainable energy budgets in birds and mammals: ecological implications. Tree 7:384-388.
    • 2 UNDESIRABLE SIDE EFFECTS OF SELECTION FOR HIGH PRODUCTION EFFICIENCY W.M. Rauw, E. Kanis, E.N. Noordhuizen-Stassen and F.J. GrommersAbstract Genetic selection has increased production levels of livestock speciesconsiderably. However, apart from a favourable increase in production, animals in apopulation that have been selected for high production efficiency seem to be more atrisk for behavioural, physiological and immunological problems. Examples are presentedof over 100 references on undesirable (cor)related effects of selection for highproduction efficiency, with respect to metabolic, reproduction and health traits, inbroilers, pigs and dairy cattle. A biological explanation for the occurrence of negativeside effects of selection is presented. Genetic selection may lead to loss of thehomeostatic balance of animals, resulting in the occurrence of pathologies andconsequently in impaired animal welfare. Future application of modern reproduction andDNA-techniques in animal breeding may increase production levels even faster than atpresent, which may result in more dramatic consequences for behavioural, physiologicaland immunological traits. Selection for more than production traits alone may preventsuch. Without knowledge about the underlying physiological processes on which geneticselection acts, selection is essentially a black box technique. Knowledge of biologicalbackgrounds will offer the opportunity to understand, anticipate and preventundesirable side effects of selection.Keywords: selection, correlated genetic response, reproduction, health, animal welfareBased on Livestock Production Science 56, W.M. Rauw, E. Kanis, E.N. Noordhuizen-Stassen andF.J. Grommers, Undesirable side effects of selection for high production efficiency in farmanimals: a review, 15-33, ©1998 Elsevier Science B.V., with permission from Elsevier Science.
    • 6 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers1. IntroductionGenetic selection has increased production levels of livestock species considerably.Since feed costs are economically the most important costs, the breeding goal in mostlivestock species is to create a population with high economic production efficiency, i.e.,high production combined with relatively low food intake (Luiting, 1990). Breeding programs have become quite successful because of the high accuracy ofbreeding value estimation, the moderate to high heritabilities of most production traitsand the use of large and fast databases containing production records of many animalsand their genetic relationships. Apart from genetic changes, production is alsoincreased by improvement of environmental factors such as housing, food composition,feeding strategies, health status and farm management. Rapidly developing techniquesin farm animal reproduction and new molecular technologies will increase in the nearfuture the intensity and accuracy of selection even more. However, apart from desired effects of genetic selection focused on higheconomic production efficiency, negative side effects have become apparent. Animals ina population that have been genetically selected for high production efficiency seem tobe more at risk for behavioural, physiological and immunological problems. This paper presents examples of undesirable (cor)related effects of selection forhigh production efficiency in broilers, pigs and dairy cows. A possible biologicalexplanation and implications for genetic selection are discussed.2. Genetic change and correlated responses to selectionSelection for high production efficiency may result in correlated responses in othertraits. The current theory about correlated selection responses is well documented intextbooks (e.g., Pirchner, 1983; Falconer and Mackay, 1996). Briefly, the observedphenotypic association (rp) between trait 1 and 2 is a function of the underlying geneticcorrelation (rg) and the environmental correlation (re), which can be written as:rp = h1h2rg + e1e2re, where ex = √(1-hx2) and hx2 is the heritability of trait x. The geneticcorrelation expresses the extent to which two characters are genetically associated;the environment is a cause of association when two characters are influenced by thesame differences of environmental conditions. It follows that the phenotypiccorrelation depends, besides genetic and non-genetic correlations between traits, onthe heritability of each trait, i.e., the proportion of the total variance which is causedby genetic variation among animals. Genetic and environmental correlations can differ inmagnitude and sign and thus the phenotypic correlation that can be observed gives noconclusion about the magnitude and sign of the genetic correlation. The correlatedresponse in trait 2 when selection is for trait 1 can be predicted by CR2 = ih1h2rgσp2,where i is the intensity of selection and σp2 represents the phenotypic standarddeviation of the correlated trait. The formula shows that the genetic correlation (rg)accounts for the sign and partly for the magnitude of the correlated selection responseand is consequently the interesting parameter when assessing genetic side effects ofselection.
    • 2. Undesirable side effects of selection for high production efficiency 7 The current animal breeding theory distinguishes two causes of genetic correlationbetween traits: linkage and pleiotropy. Linkage is the situation in which different lociinfluencing separate traits are situated close together on the same chromosome,preventing the genes from segregating independently at meiosis. Pleiotropy is thesituation where a single gene affects two or more different traits. Production traitsare usually composed of many underlying synchronously proceeding co-operativemetabolic processes, each of which is more or less genetically determined. Thebiochemical reactions and control mechanisms on which a gene and thus geneticselection acts, are likely to influence more than one trait. The actual geneticcorrelation between two traits is the net effect of pleiotropy and linkage. Falconer andMackay (1996) state that estimates of genetic correlations are usually subject torather large sampling errors and are therefore seldom very precise; moreover they maydiffer in different populations since the genetic correlations are strongly influenced byallele frequencies. Genetic side-effects of selection may be assessed from selection experiments orfrom estimated genetic relationships between the traits of interest. Data obtainedfrom selection experiments are mostly on experimental animals kept under relativelysimilar and constant conditions which consequently reduces the environmental variation.Most selection experiments include a control line and/or two divergently selected linesin order to separate correlated genetic responses from environmental changes.However, due to long generation intervals and high experimental costs in largerlivestock species, observations may be on few animals only, selected for relatively fewgenerations. In those species, most information is based on field data and thus subjectto management policy, herd size, season, age, breed, etc. (Grommers, 1987). Advancedstatistical techniques attempt to account for these environmental effects, but theyestimate genetic parameters under the assumption of a simplified linear model whichcan produce no more than an approximation of the true state of nature. In practice, realised heritabilities for traits under selection in commercial broilers,pigs and dairy cows seem to remain positive, resulting in a continuous positive selectionresponse for production traits. This gives, however, no conclusion about the futurestate of the animal as will become clear from the reviewed side-effects of selection:e.g., selection for high body weight in turkeys has led to a continuous increase in bodyweight, but also resulted in extremely heavy male animals that make natural matingimpossible. Several long term selection experiments, especially in mice, did show limits toselection (Eisen, 1980; Barria and Bradford, 1981; Bünger et al., 1984; Canon et al.,1992), supporting the statement of Falconer and Mackay (1996) that response ofselection can not be expected to continue indefinitely. Several reasons have beenpostulated why limits to selection for traits under selection have not yet been reachedin commercial livestock species as compared with experimental results. Commercialselection has been less intense than that applied in selection experiments. Moreover, anaggregate genotype consisting of several traits will prolong the expected period ofresponse (Fredeen, 1984; Hunton, 1984). Improved management and quality of theenvironment reduces environmental variability allowing for relatively greater expressionof genetic variability, and thus increased estimated heritabilities.
    • 8 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers Application of modern techniques will alter the speed at which genetic increase inproduction is achieved. Rapidly developing techniques and commercial use of in vitromaturation and fertilisation, gene transfer, sperm and embryo sexing and embryocloning will increase the intensity of selection. New molecular technologies, allowing thelivestock breeders to select on genotype rather than on phenotype, will increase theefficiency of selection in farm animals and the rate at which genetic gains will beachieved even more (Hansel and Godke, 1992). Gene mapping and marker assistedselection will enable selection early in life unaffected by micro-environmental variation,equally in both sexes, and without requiring costly trait evaluation, which will increaseintensity and accuracy of selection and decrease generation interval (Soller, 1994).However, it will also increase the speed of expression of detrimental antagonisticrelationships with selection for high production efficiency. In order to anticipate undesirable genetic correlated responses to selection, thosehave to be described first. In this review, examples in the literature on undesirablegenetic side effects of selection and on genetic correlations between production traitsand metabolic, reproduction and health traits will be discussed. Three livestock speciesare considered with one product each: meat type poultry, meat type pigs and dairycattle.3. Undesirable side-effects of selection for high production efficiency3.1. Broilers and turkeys3.1.1. Responses in production traitsThe primary aim in poultry meat production is rapid growth, superior food efficiencyand high processing yield. Poultry breeding animals have shown a continuous response toselection for growth rate with heritabilities of approximately 40% (Cahaner and Siegel,1986). Moreover, application of quantitative genetic and biometrical methods to poultrybreeding has resulted in a genetic increase in production efficiency (Reddy, 1996).About 85% to 90% of the increase in body weight resulted from genetic selection forincreased body weight (Havenstein et al., 1994a). Table 1 shows the increase in production in broilers and turkeys from 1960 to1996. Age at a given body weight in broiler chickens is reduced (Cahaner and Siegel,1986) by about 1 day per year (Havenstein et al., 1994a).3.1.2. Responses in metabolic traitsThe genetic increase in production has been supported by a corresponding correlatedgenetic increase in daily food consumption (Dunnington and Siegel, 1995) and foodefficiency (Owens et al., 1971; Marks, 1979; Barbato et al., 1984; Havenstein et al.,1994a; Dunnington and Siegel, 1995). It appears that in general most of the heritablevariation in body weight gain is associated with differences in food intake (McCarthyand Siegel, 1983). Dunnington and Siegel (1996) reported that superior food efficiency
    • 2. Undesirable side effects of selection for high production efficiency 9of a high body weight line over a low body weight line, in a divergent selectionexperiment for body weight in broiler chickens at eight weeks of age, was associatedwith several physiological factors, including a decrease in metabolic rate as measuredby oxygen consumption (Owens et al., 1971), a higher rate of food passage and digestion(Cherry and Siegel, 1978; Dunnington and Siegel, 1995) and higher enzymatic activitiesin the small intestine (Dunnington and Siegel, 1995). Lepore et al. (1963) observed amore efficient utilisation of energy and the amino acid histidine in embryos of the highbody weight line compared with the low body weight line.Table 1 Means of age at slaughter (AGE), liveweight at slaughter (SW), food conversion (FC) andgrowth rate (GR) in broiler chickens (B) and turkeys (T) from 1960 to 1996.Trait 1960 1965 1970 1975 1980 1985 1990 1995 1996 bAGE (B) (d) (NL) 59 51 48 49 46 45 43 44 bSW (B) (g) (NL) 1250 1330 1420 1560 1630 1790 1860 1900SWc (B) (g) (USA) 1680 1745 1820 1880 1975 2095 2185 2330SWd (T) (kg) (USA) 7.53 8.21 8.92 8.88 9.26 10.03 10.63 11.58 a b b b b a,b bFC (B) 2.5 2.32 2.10 2.03 2.04 2.0 1.92 1.83b 1.83bGR (B) (g/d) 10a 21b 26b 30b 33b 37b, 40a 42b 44b 45ba Cahaner and Siegel (1986); bLEI-DLO (1996); cUSDA (1995a); dUSDA (1995b). Both the high and the low body weight line differed for several traits related toappetite (Dunnington and Siegel, 1996). Experimental lesions made in the hypothalamusof hens from both lines resulted in an expected obesity syndrome in the low bodyweight line, but not in the high body weight line. This suggests that selection forincreased body weight damaged the hypothalamic satiety mechanisms leading to afailure to diminish the hunger drive and consequently to hyperphagia oroverconsumption (Burkhart et al., 1983). These results are supported by an experimentusing the force feeding technique, which forces chickens to eat more than their adlibitum food intake: chickens from the low body weight line could be fed substantiallyabove ad libitum food intake, whereas in the high body weight line this was possible to asignificantly lesser extent. At day one of age, chickens of the high body weight linecould not be overfed, whereas the low body weight line could be fed 150% above adlibitum; during the first three weeks of age the high body weight line and the low bodyweight line could be fed 10% and 23% above ad libitum food intake, respectively. Theresults suggest that chickens selected for high body weight are inclined to eat a volumeof food corresponding to the size of their gastro-intestinal capacity: they consumefood above their metabolical requirements until they reach a limit set by their gastro-intestinal capacity (Barbato et al., 1984; Nir et al., 1978). Denbow et al. (1986) showedincreased appetite in a high body weight line compared with a low body weight line as aresult of increased responsiveness to biogenic amines, which regulate food intake. Intense selection pressure for body weight in broilers has altered the growthpattern so that birds reach slaughter weights at younger ages (Marks, 1979; Anthonyet al., 1991). High body weight line chickens grew faster and had a higher percentage ofbody fat than the low body weight line (Chambers et al., 1981; Barbato et al., 1984;Dunnington and Siegel, 1996) and compared with a randombred line (Havenstein et al.,
    • 10 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers1994b). The higher percentage of body fat in the high body weight line resulted mainlyfrom decreased rates of lipolysis (Calabotta et al., 1985). It is suggested thatexcessive fat deposition in chickens selected for rapid growth is associated withincreased concentrations of insulin and glucagon in plasma and perhaps insulinresistance (Sinsigalli et al., 1987).3.1.3. Responses in reproduction traitsSelection for fast growth rate has resulted in correlated negative responses forseveral reproduction traits. A dramatic consequence of long-term selection forincreased muscle in turkeys is the damage made to the back of the hens by the heavymales with natural mating. One hundred percent artificial insemination allowed for thecontinuation of intense selection for body weight in male lines (Hunton, 1984; Cahanerand Siegel, 1986; Dunnington, 1990). Fertility is reduced in broiler breeder hens withexcessive body weight (Dunnington, 1990; Liu et al., 1995). It is generally accepted thatreproductive problems are common in broiler strains, and husbandry practices areroutinely employed to reduce adult body weight (McCarthy and Siegel, 1983). Broiler breeders selected for high body weight produced a higher number of eggsthan broilers selected for low body weight, but a higher percentage of defective eggs(e.g., double yolk, extra-calcified, shell-less, soft shelled): 17.1% and 2.4% in the highand the low body weight line, respectively (Van Middelkoop and Siegel, 1976; Anthonyet al., 1989). This increase in number of defective eggs produced was in large part dueto lost synchrony of ovulation and packaging of the eggs (Dunnington and Siegel, 1996).The high body weight line showed an increased ‘erratic ovulations and defective eggsyndrome’ (EODES) quantified by the ratio of percentage daily production of normaleggs (DP) over percentage daily ovulation (DO) (Jaap and Muir, 1968; Siegel andDunnington, 1985; Liu et al., 1995). Ratios of DP:DO were for the high and the low bodyweight line 0.84 and 0.98, respectively (Anthony et al., 1989; Liu et al., 1995).Percentage hatch of fertile eggs decreased significantly with 1.11% per generation inbroilers selected for 15 generations for high 12-week body weight (Maloney et al.,1967). Embryos of a high 8-week weight selected broiler line revealed a higherfrequency of chromosomal abnormalities (14.5%) than a line selected for low bodyweight (6.2%) (Reddy and Siegel, 1977). Inconsistent results have been reported on the effects of selection for high bodyweight on several semen characteristics. Marini and Goodman (1969) found a lowersemen concentration, volume and motility and higher percentage dead and abnormalsperm cells in a high 12-week weight selected line compared with a low weight selectedline of broilers in generation 15. In generation 18, however, only sperm concentrationwas reported to be significantly different (Cheng and Goodman, 1976). Nestor (1977)observed a lower semen concentration and volume in 16-week body weight selectedturkeys compared with a randombred control line. Siegel (1963) observed a highersemen volume in high 8-week weight selected broilers compared with low weightselected broilers; motility was lower and no differences in semen concentration werefound. Edens et al. (1973) showed that spermatozoa of a high body weight line possessa lower potential for a sustained endogenous metabolism than the spermatozoa of a low
    • 2. Undesirable side effects of selection for high production efficiency 11body weight line, which may be related to the lower sperm motility found in the highbody weight line.3.1.4. Responses in health traitsSelection for high body weight has resulted in a correlated negative immuneperformance. Broilers selected for high growth rate showed lower antibody responseswhen challenged with sheep erythrocytes (SRBC) than a low body weight line (Miller etal., 1992) and a randombred control line (Qureshi and Havenstein, 1994). Little or nodifferences were found between lines in macrophage and natural killer cell functions,i.e., the non-adaptive components of the immune system (Qureshi and Havenstein,1994). Nestor et al. (1996a,b) found a significantly higher percentage of mortality inturkeys selected for high 16-week body weight compared with a randombred controlline in a natural outbreak of erysipelas, 11.8% and 1.6%, respectively, and whenchallenged with either Pasteurella multocida, 72.1% and 43.6%, respectively, andNewcastle disease virus, 32.5% and 15.8%, respectively. Havenstein et al. (1994a) reported a more than four times higher mortality at 42days of age in a commercial 1991 broiler strain (9.7%) as compared with a randombredbaseline originating from 1957 (2.2%). The lines used in this experiment did notoriginate from the same population. Most of the mortality in the commercial strainafter three weeks of age was associated with flipovers (sudden death), ascites and legproblems. Moreover, incidence of tibial dyschondroplasia was 47.5% in the commercialline compared with 1.2% in the randombred baseline. The lines differed greatly inseverity of the lesions: in the commercial line, 42.4% of the lesions involved more thanone-fourth of the tibial head, whereas in the randombred line all lesions involved lessthan one-fourth of the tibial head. Leenstra (1993) indicated that male chickens from aline selected for high body weight at 6 weeks of age showed more cases of severedyschondroplasia (28%) than male chickens from a line selected for low food conversion(3%). Moreover, broilers selected for high body weight had a higher mortality rate(7.4%) than broilers selected for low food conversion (2.0%) which was mainly causedby infectious diseases and heart and circulation problems. Ascites, i.e., the accumulation of edematous fluid within the abdominal cavity, is adisease found at high altitutes under hypoxic conditions. Since approximately 1980ascites in broiler chickens is also observed at lower altitudes, even at sea level. In TheNetherlands, ascites, as a cause of mortality in broiler chickens, is increasing steadily,running parallel with a faster growth rate and, as a result, an increasing metabolic rate(Scheele, 1996). A high incidence of heart failure syndrome and ascites was found infast growing broilers exhibiting a low food conversion ratio resulting from low values ofheat production per gram body weight gain and oxygen consumption per gram depositedprotein (Scheele, 1996). Heart failure syndrome and ascites may be initiated by alimited thyroid hormone production of which the main effect is to increase metabolicrate by stimulating oxidative metabolism (Scheele et al., 1992). Havenstein et al.(1994b) and Dunnington and Siegel (1995) observed a decreased heart and lung size asa percentage of body weight with genetic selection for increased growth rate. It issuggested that faster growth in broiler chickens requires more oxygen to support the
    • 12 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommershigher metabolic rate. If oxygen becomes a limiting factor, the same diseases mayoccur as are known under hypoxic conditions, like heart failure and ascites (Scheele,1996). The metabolic properties of a muscle are related to the ability of an animal to copewith environmental stresses: muscles with a high endurance capacity, low proportion ofwhite glycolytic fibres, high capillary supply, high oxidative capacity and smaller fibresare better in conserving their energy than muscles with the opposite characteristicsand consequently better able to sustain environmental stresses (Henckel, 1992). Itappears that in pigs, the glycogen level in muscles may be more related to PSE (pale,soft and exudative meat) proneness, which is preceded by PSS (Porcine StressSyndrome), than halothane sensitivity (Henckel et al., 1992). Henckel (1992) observed asurprisingly low proportion of oxidative slow twitch muscle fibres and a high proportionglycolytic fast twitch fibres in different muscles of chickens. This suggests that mostcommercial chickens may be very sensitive to environmental stresses, resembling thesituation observed in pigs. Indeed those muscles in the chicken would be classified asPSE (Henckel, 1992) as has been observed in turkeys (Barbut, 1997). Henckel (1992)suggests that since glycolytic fast twitch muscle fibres have higher growth potentialsthan other fibre types, the increase of glycolytic fast twitch fibres may be a result ofselection for high growth rate. Turkeys selected for high 16-week body weight required fewer inductions to reacha longer duration of tonic immobility than birds from a randombred control line implyinga higher state of fearfulness in the high body weight line (Nestor et al., 1996a).3.2. Pigs3.2.1. Responses in production traitsIn pig production, selection is mainly on high growth rate and/or minimum backfatthickness, i.e., high lean tissue growth rate, on low food conversion, soundness andrecently also litter size. A continuous increase in growth rate and decrease in foodconversion from 1960 to 1996 is shown in Table 2 in Dutch and Norwegian meat typepigs. Dutch figures originate from on farm testing; Norwegian figures originate fromsib-testing on sisters and castrates on testing stations. Because in breedingprogrammes much emphasis is placed on production traits, and heritabilities aremoderate (growth rate, food conversion) to high (meat percentage), the increase inproduction efficiency is probably to a high extent of genetic origin.Table 2 Means of growth and food conversion (FC) in Dutch (NL) and Norwegian (N) meat typepigs from 1960 to 1996.Trait 1960 1970 1974 1977 1980 1983 1986 1989 1992 1995 1996 aGrowth (g/d) (NL) 606 611 646 651 693 718 717 729 bGrowth (g/d) (N) 629 706 835 888 897 936 960 956 962FC (NL)a 3.39 3.27 3.15 3.08 2.98 2.93 2.87 2.79FC (N)b 3.24 2.83 2.54 2.50 2.45 2.31 2.26a b IKC (1996); Norsvin (1996).
    • 2. Undesirable side effects of selection for high production efficiency 133.2.2. Responses in reproduction traitsUnder normal conditions a sow shows oestrus between 4 and 7 days after weaning, butthe time from weaning to oestrus is highly variable, especially in first-litter sows, andmay be 200 days or longer (Ten Napel, 1996). Analyses of data from pigs selected forhigh daily gain and low backfat within two farms showed no clear genetic relationshipwith incidence of prolonged intervals from weaning to farrowing when only sows with anobserved interval were included. However, when sows that were culled after normallyweaning a litter were supposed to be culled for anoestrus, and were included into theanalyses as having a prolonged interval, genetic relationships between rebreedingperformance and the traits under selection were consistently unfavourable and mostlysignificant. Large White pigs with a prolonged interval from weaning to farrowing had a14.6 g/d higher breeding value for average daily gain and a 0.64 mm lower breedingvalue for backfat adjusted for body weight (Ten Napel and Johnson, 1997). Genotypedoes not affect the weaning-to-oestrus interval in a direct way, but acts in an indirectway through genetic variation in susceptibility to factors prolonging ‘interval weaning tostart of cycle’. It is stated that genetic variation in susceptibility to severe depletionof body reserves and stressors may explain the observed genetic variation in weaning-to-oestrus interval (Ten Napel et al., 1995).Table 3 (a) Genetic correlations between daily growth rate, lean percentage and backfat, andseveral reproduction traits in pigs; (b) Number of animals (N), breed, average growth rate, leanpercentage and backfat.a Trait Daily growth rate Lean percentage Backfat c c Intensity vulvar symptoms 0.19 -0.17 Duration pro-oestrus (d) 0.03c -0.09c Standing reflex -0.61c 0.10c Duration standing oestrus (d) -0.49c 0.02c a b c Age at puberty (d) -0.38 , -0.31 , -0.07 0.20b, 0.40c 0.27a Age at farrowing -0.61d -0.16d Piglet weight 0.50a, 0.41b 0.14b 0.19ab N Breed Average Growth rate Average Lean % Average backfat (g/d) (mm) a 737 D, Y, S, LR gilts 690 (weaning - 90.7 kg) 25.2 (90.7 kg) b 393 Y gilts 823 (25 - 90 kg) 58.5 (90 kg) c 740 Y gilts 832 (25 - 90 kg) 58.2 (90 kg) d 4068 Y sows 525 (birth - 80 kg) 11.9 (80 kg)a b c d Hutchens et al. (1981); Rydhmer et al. (1992); Rydhmer et al. (1994); Rydhmer et al. (1995);D = Duroc; Y = Yorkshire; S = Spot; LR = Landrace. Genetic correlations between daily growth rate, lean percentage and backfat, andseveral reproduction traits are presented in Table 3. Hutchens et al. (1981) definedpuberty as first detected oestrus indicated by a standing response to a boar; in thetwo other studies, puberty was defined as the first ovulation judged by progesteronelevels. Pro-oestrus was defined as the period before standing oestrus when reddening
    • 14 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommersand swelling of the vulva was observed, and length of standing oestrus was defined asthe number of days that the gilt showed a standing reflex. Gilts with higher leanpercentage had genetically delayed onset of puberty, showed shorter pro-oestrus, lessintense and shorter reddening and swelling of the vulva at puberty, and had a higherability to show standing reflex. Kirkwood and Aherne (1985) suggest that selection forleanness is also selection for a larger mature size and, on a chronological time scale, aphysiologically younger animal. Selection for increased growth rate, efficiency andleanness appears to have resulted in an increase of some 30% in mature size over 20years (Whittemore, 1994). If a minimum percentage body fat is required for the onsetof oestrus, puberty can be expected to be delayed. However, Hutchens et al. (1981)found a positive genetic correlation between body weight adjusted backfat (which isinversely related to leanness) and age at puberty. Growth rate was genetically negatively correlated with the ability to show standingreflex, duration of standing oestrus and onset of puberty, and positively with theintensity of vulvar symptoms. When genetic correlations between growth rate andoestrus traits and leanness and oestrus traits have opposite signs, the consequence ofselection for high lean tissue growth rate on oestrus traits will depend on the relationbetween growth rate and leanness in the breeding goal (Rydhmer et al., 1994, 1995). Piglet weight, which is positively related to piglet survival, growth of slaughter pigsand litter size of breeding sows, was positively correlated with both growth rate andlean percentage (Rydhmer et al., 1992). However, Hutchens et al. (1981) observed apositive genetic correlation between backfat and piglet weight. Kerr and Cameron(1995) observed that selection for lean growth did not significantly affectreproductive performance, but that animals selected for high lean food conversion orlow daily food intake had reduced reproductive performance: litter size and weight atbirth were reduced in the high lean food conversion line (9.1 vs. 10.7 kg) and the lowdaily food intake line (9.6 vs. 10.2 kg) compared with their corresponding control lines(10.7 vs. 13.7 kg and 11.1 vs. 13.7 kg, respectively). Between-study variation in sign and magnitude of the genetic correlations betweenproduction traits and reproduction traits may result from the fact that the parameterestimates are experiment specific and from differences in selection strategies,populations and environments within the studies (Kerr and Cameron, 1996). Moreover,trait definitions may vary between studies (Rydhmer et al., 1992).3.2.3. Responses in health traitsLacombe boars selected for high lean tissue growth rate in an experiment of Sather(1987) showed significantly more leg weakness in the foreleg (29.5%) and rear leg(41.5%) than non-selected control boars (18.4% and 28.1%, respectively). Average dailygrowth rate from 56 days of age to 90 kg body weight was 861 and 777 g, and averagebackfat adjusted to 90 kg body weight 15.6 and 16.3 mm for selection line and controlline boars, respectively. Effects in gilts were smaller and only significant for forelegscores (9.2% and 3.6% for selected and control line gilts, respectively). Frond-endstructural leg soundness scores in barrows and gilts were not significantly affected byselection in the 5th generation of a divergent selection experiment for postweaning
    • 2. Undesirable side effects of selection for high production efficiency 15average daily body weight gain (Woltmann et al., 1995). The selection differential fordivergence was 470 g/d or approximately five standard deviations. Leg weakness can be to some extend explained by the occurrence ofosteochondrosis. Table 4 shows genetic correlations between production traits and legweakness and osteochondrosis (OC) scores. Pigs with high leanness and growth rate hadworse leg scores, and worse OC scores in both the elbow and the knee joint (Lundeheim,1987; Huang et al., 1995). Webb et al. (1983) found in a group of 23 975 Large Whiteand Landrace boars adverse genetic correlations of -0.20 to -0.40 between fat depthmeasurements and aggregate leg scores which represented both the number andseverity of several leg problems.Table 4 (a) Genetic correlations between lean percentage and growth rate, and leg weaknessscore and osteochondrosis score for elbow and knee. (b) Number of animals (N), breed, averagegrowth rate and lean percentage.a Trait Lean percentage Daily growth rate d a b Leg weakness score (0-5) -0.43 , -0.09 -0.26a, -0.35b, -0.39c Osteochondrosis elbow score (0-5)e 0.22a, 0.17b 0.10a, 0.19b Osteochondrosis knee score (0-5)e 0.28a, 0.08b 0.29a, 0.03bb N Breed Average growth rate (g/d) Average lean percentagea 5568 LR 883 (30 - 100 kg) 59.1 (100 kg)b 4318 Y 897 (30 - 100 kg) 59.1 (100 kg)c 2257 LR, Y, D 888 (30 - 110 kg)a, b c d e Lundeheim (1987); Huang et al. (1995). 0 = worst, 5 = best; 0 = best, 5 = worst;Y = Yorkshire; LR = Landrace; D = Duroc. More oxidative intermediate and oxidative slow twitch muscle fibres have beenfound in the Longissimus dorsi muscle in wild pigs compared with domesticated pigbreeds by Rahelic and Puac (1980) and Essén-Gustavsson and Lindholm (1984),respectively. Moreover, oxidative capacity was significantly higher and glycogen contentsignificantly lower in wild boars. They had significantly larger mean fibre areas, alarger number of capillaries per fibre, and higher oxidative enzyme activities(Karlström, 1995). These results may indicate that domestication of pigs has resulted ina decreased ability to sustain environmental stresses. Investigation into direct effectsof selection for high growth rate on muscle composition is desired.3.3. Dairy cattle3.3.1. Responses in production traitsIn most dairy cattle breeding programs, selection is mainly for high milk yield. Table 5shows the continuing increase in (daily) milk production that has been achieved in thepast decades. In dairy cow populations heritabilities of milk production are foundaround 0.20 to 0.35 (Kennedy, 1984).
    • 16 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers3.3.2. Responses in metabolic traitsIn early lactation, high producing cows are generally in negative energy balance (i.e.,nutrient intake minus requirements) and mobilise body reserves for milk production(Freeman, 1986; Butler and Smith, 1989). Negative energy balance generally reaches itsmaximum during the first two weeks of lactation and recovers at a variable rate (Butlerand Smith, 1989). Harrison et al. (1990) found a significantly lower energy balance in 10selected cows than in 10 non-selected cows at 1, 2, 10, and 11 weeks postpartum, in a20-year selection experiment for high Predicted Differences in milk yield; 305-daymature equivalent milk yield was 10 814 kg in the high and 6912 kg in the average line.Table 5 Means of milk production per lactation (MP) and days in milk per lactation (DIM) in TheNetherlands (NL), Norway (N) and United States (USA) from 1950 to 1996.Trait 1950 1955 1960 1965 1970 1975 1980 aMP (kg) (NL) 4029 4118 4372 4370 4639 5063 5466 bMP (kg) (N) 2932 3723 4919 5428 5750 cMP (kg) (USA) 4017 4874 5180 5938DIM (d) (NL)a 308 308 306 304 305 309 311 1985 1987 1989 1991 1993 1996 aMP (kg) (NL) 5559 6214 6687 7001 7220 7605 bMP (kg) (N) 5716 6212 6261 6264 6403 6265 cMP (kg) (USA) 6516 6893 7122 7434DIM (d) (NL)a 304 306 311 314 322 325a b c NRS (1996); NML/Norske Meierier (1996); USDA (1991). Body condition scoring, based on a scale from 1 (thin) to 5 (obese), is an acceptedsubjective method to estimate the degree of fatness to assess body reservesregardless of frame size and body weight (Gallo et al., 1996). Veerkamp et al. (1994)and Gallo et al. (1996) reported an inverse relationship between milk yield and bodycondition score. Mean body condition score of 219 Holstein cows bred from bulls withhighest genetic merit for fat + protein was significantly lower than for 158 non-selected control cows (2.39 vs. 2.54) in a selection experiment of Veerkamp et al.(1994); average 26-week milk yield for the selection and control line was 5540 and5007 kg, respectively. Loss of body condition score during early lactation was almosttwice as high for 401 Holstein cows yielding more than 12 000 kg as for 187 cowsyielding less than 6000 kg milk (-0.64 vs. -0.38 units) (Gallo et al., 1996). A negative energy balance may be associated with a higher incidence of metabolicdisorders, impaired fertility, and other health problems. However, small sample size andthe lack of power to detect significant associations may account for controversialresults found between authors. Results on relationships between milk production andseveral metabolical traits are mainly of phenotypic nature since selection experimentswith dairy cows are rare and field data on metabolical traits are practically notavailable.
    • 2. Undesirable side effects of selection for high production efficiency 17 Timing, magnitude and recovery rate of negative energy balance may be related toreinitiation of the ovarian activity and interfere with the ability of the hypothalamo-hypophyseal axis to regulate the luteinizing hormone (LH) pulse frequency necessaryfor ovarian follicular development and ovulation (Butler and Smith, 1989; Canfield andButler, 1990), and with the ovarian responsiveness to LH signalling (Canfield and Butler,1991). First ovulation occurred on average 10 days after the most negative energybalance (Butler et al., 1981). The exact way in which negative energy balance maymodulate LH secretion is not known. Non-esterified fatty acids provide a potentialsignal of status of energy balance to neural centres controlling LH secretion (Canfieldand Butler, 1991). Failure of early reinitiation of ovarian activity may result in fewerovulatory cycles before insemination and thus result in a lower conception rate anddecreased fertility (Butler and Smith, 1989; Senatore et al., 1996). Harrison et al.(1989) suggest that suppression of oestrus behaviour, rather than the time to initiationof ovarian activity in high-producing Holstein cows may be a factor affecting theinterval from parturition to conception. Barnes et al. (1985) found in a selection experiment similar insulin concentrations,higher mean growth hormone (GH) concentrations and lower mean glucagonconcentrations in 18 Holstein cows selected for high milk production compared with 18cows from a randombred control line. It is suggested that this combination is related toincreased conversion of adipose tissue into energy source, reduced lipogenesis, andincreased efficiency of lipolysis in the selected line. Average mature equivalent305-day milk production was 7890 kg in the selection group and 6885 kg in the controlgroup. Bonczek et al. (1988) reported a decreased insulin concentration, an increasedsomatotropin concentration and unchanged concentrations of prolactin and thyroxinewith selection for milk yield: insulin concentrations were 18.80 and 22.50 µIU/ml, andsomatotropin concentrations were 4.46 and 3.73 ng/ml in 29 Holstein cows selected formilk yield and 23 unselected cows, respectively. 305-d milk yield averaged 9878 kg forthe selection line and 7402 kg for the control line. Insulin may be a signal to the ovaryof metabolic recovery from pregnancy and lactation (Canfield and Butler, 1991).3.3.3. Responses in reproduction traitsAntagonistic relationships between high milk production and several fertility traitshave been observed by several authors (e.g., Hansen et al., 1983; Hoekstra et al. 1994);others, however, found no relationship (e.g., Villa-Godoy et al., 1988; Raheja et al.,1989). Nebel and McGilliard (1993) pointed out in a review about interactions betweenhigh milk yield and reproduction, that a trend can be observed in associations foundbetween the two traits. Data collected prior to 1970 show little or no association,whereas an antagonistic phenotypic relationship has been reported more frequently,with increasing milk production, after 1975. Although few, some selection experiments have been conducted to investigate theconsequences of selection for high milk production on reproductive performance;results are given in Table 6. Table 7 gives genetic correlations of 305-day milkproduction with several fertility traits as analysed by several authors. Geneticcorrelations of 60-, 80-, and 100-day milk yield with fertility were quite similar in sign
    • 18 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommersand size compared with 305-day milk yield (Berger et al., 1981; Van Arendonk et al.,1989). Results show that in general high producing cows were bred later, showed moredays open, had a longer calving interval, a lower rate of non-return at 56 days, andrequired more services per conception than low producing cows. Ducker and Morant(1984) suggest that both production level and rate of increase in yield in early lactationare associated with reproductive performance. Long-term experiments in dairy cows have not been conducted due to the longgeneration interval and high maintenance costs (Kennedy, 1984; Legates and Myers,1988). Most information is based on field data and thus subjected to aforementionedmanagement policy and possible preferential treatment of high producers. Most datasets contain records that include only cows conceiving and calving normally (Berger etal., 1981). Butler and Smith (1989) and Nebel and McGilliard (1993) state that theseveral interval traits (e.g., days open, calving interval, etc.) are too much dependent onmanagement policy. Estimated correlations can be affected if high yielding cows arebetter detected for oestrus, are inseminated later (Berger et al., 1981), are given moreopportunities for reinsemination than average and low yielding cows (Raheja et al., 1989;Hoekstra et al., 1994) or are culled less frequently (Fonseca et al., 1983; Eicker et al.,1996). Conception rate of first breeding or number of breedings are thought to bebetter indicators of reproductive function (Nebel and McGilliard, 1993; Hoekstra et al.,1994).3.3.4. Responses in health traitsMuch controversy exists about the relationship between production and health traits.Shanks et al. (1978) observed in 171 cows selected for high genetic potential for milkproduction 9% more cases of digestive disorders (20% and 11%, respectively), 5% morecases of foot rot (9% and 4%, respectively), 14% more cases of skin or skeletaldisorders (49% and 35%, respectively), 11% more cases of udder edema (i.e., swelling inthe mammary and adjacent tissue), and 2% more lactations affected by mastitis than187 cows selected for low genetic potential. No differences were found between thegroups for respiratory disorders. Cows were subsequently randomly assigned to bebred to bulls with either high or low average Predicted Difference in milk production.Results after the first 6.5 years of selection showed 5% more joint or leg injuries inthe selected line (7%) compared with the randombred control line (2%), 3% fewermammary cuts (0% and 3%, respectively), 13% more total cases of skin or skeletaldisorders (54% and 41%, respectively), and 19% more cases of udder edema. Nodifferences were found between the lines in digestive disorders or respiratorydisorders. Wautlet et al. (1990) found in a 23-year selection experiment for high milkyield, comparing 139 selected cows with 172 non-selected control cows, no significantdifferences in edema, dystocia or retained placenta. Health registering in the Nordic countries provides valuable data for investigatingthe epidemiological and genetic background of several health traits. Table 8 givesgenetic correlations of 305-day milk production with several health traits as analysedby several authors. These results, although in some cases controversial, suggest that
    • Table 6 Differences between cows selected for high milk yield (HY) and control (C) line cows for postpartum (PP) interval to first ovulation,postpartum interval to first oestrus, postpartum interval to first service, number of inseminations, and days open as analysed by several authors (A). Production level Observations A D (yr) Breed N HY C Trait HY C 1 5 16 H 625 9100 kg 6883 kg Differences in trend over 305-d MP 305-d MP years for days open NS between groups 2 20 H 206 10,814 kg 6912 kg PP interval to 1st ovulation (d) 31 29 NS 305-d MP 305-d MP PP interval to 1st oestrus (d) 66 43 ** Days open (d) 217 74 ** 3 5 17 J 1056 6594 kg 5528 kg PP interval to 1st oestrus (d) 32 30 NS 1st lact 1st lact PP interval to 1st service (d) 88 77 *** Days open (d) 110 99 ** 4, 8 7 19 H 187 5454 4944 PP interval to 1st oestrus (d) 45 49 NS 26-wk MP 26-wk MP Number of inseminations (d) 1.9 1.8 NS1 2 3 4 5 6Legates and Myers (1988); Harrison et al. (1990); Bonczek et al. (1992); McGowan et al. (1996); Whole selection experiment; Last generation;7 Last four calving seasons; 8Selection is on genetic merit for kg fat + protein; D = duration of selection experiment; H = Holstein; J = Jersey;MP = milk production; N = number of observations; **P<0.01; ***P<0.001; NS = not significant. 2. Undesirable side effects of selection for high production efficiency 19
    • Table 7 (a) Genetic correlations between 305-d milk yield (305-MY) and days (from calving) to first insemination (FI), days open (DO), number of 20services per conception (NS), 56 non-return rate after first insemination (NR56) and calving interval (CI), for parity (PAR) 1, 2 and 3 in dairy cows;(b) Average 305-d milk production (kg) (AP), total number of records (N) and breed.a b PAR 1 PAR 2 PAR 3 PAR 305-MY AP N AP N AP N a* b d e f g a*FI 1 0.48 , 0.25 , 0.01 , 0.22 , 0.44 , 0.48 6017 - 6837 - 7343 - a* b e b 2 0.32 , 0.34 , 0.23 - 41 710 - 31 162 - 22 389 c 3 0.49a*, 0.27b, -0.08e 7249 3976 dDO 1 0.62a*, 0.37b, 0.54c, 0.68d, 0.64e 4374 5010 a* b e e 2 0.15 , 0.43 , 0.65 5113 6216 6108 4815 6518 3541 a* b e f 3 0.18 , 0.39 , 0.27 6197 82 659 gNS 1 0.62a*, 0.12b, 0.67d 5765 9163 a* b e 2 0.53 , 0.42 , 0.46 3 0.10a*, 0.25b, 0.36eNR56 1 -0.26f Breed 2 -0.56e a e 3 -0.28e H DF d f g b fCI 1 0.66 , 0.55 , 0.32 - H/DF c g 2 H H d 3 SBa Berger et al. (1981); bHansen et al. (1983); cSeykora and McDaniel (1983); dSchneeberger and Hagger (1986); eVan Arendonk et al. (1989); fHoekstraet al. (1994); gPryce et al. (1997); H = Holstein; SB = Swiss Braunvieh; DF = Dutch Friesian; *305-day fat corrected milk yield. W.M. Rauw, E. Kanis, E.N. Noordhuizen-Stassen and F.J. Grommers
    • Table 8 (a) Genetic (rg) and environmental (re) correlations between 305-day milk yield (305-MY) and mastitis (MS), ovarian cyst (OC), ketosis (KT),milk fever (MF), displaced abomasum (DA), retained placenta (RP), and leg problems (LP) for parity 1 (P1) or across parities (ACP) in dairy cows;(b) Breed, average 305-day milk production (kg) (AP) and total number of records (N).a 305-MY b Breed P1 ACP rg re AP N AP N a d f g h i 7 aMS P1 0.21 , 0.39 , 0.51 , 0.37 , 0.46 , -0.40 -0.01 - 4400 890 b d e i b ACP 0.26 , 0.57 , 0.18 , 0.21 H 8924 5091 cOC P1 -0.14g 0.277 FA - - - 70 775 d ACP -0.01e, -0.06g 0.207 FA - - - 70 775 c f eKT P1 0.30 , 0.65 H - 11 008 c e g 7 f ACP 0.10 , 0.26 , 0.77 0.02 - 5017 216 565 gMF P1 H - - - 7416 b e g 7 h ACP 0.27 , 0.33 , -0.67 0.15 FA 5334 10 152 iDA P1 H 5765 9163 6455 33 732 e g 7 ACP -0.15 , -0.04 -0.10RP P1 ACP 0.26b, -0.43eLP P1 0.24i ACP 0.32e, 0.27g, 0.29i -0.187a Bunch et al. (1984); bThompson (1984); cGröhn et al. (1986); dSyväjärvi et al. (1986); eLyons et al. (1991); fSimianer et al. (1991); gUribe et al. (1995);h Pösö and Mäntysaari (1996); iPryce et al. (1997); H = Holstein; FA = Finnish Ayrshire. 2. Undesirable side effects of selection for high production efficiency 21
    • 22 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommersgenetic increase in milk yield leads in general to cows higher at risk of mastitis, ketosisand leg problems, but with a lower occurrence of ovarian cyst and displaced abomasum. As for reproductive disorders, field data on health traits are confounded withherd and management factors: e.g., the ability of the dairyman to recognise a disease,preferential treatment of high producers, different nutritional treatments betweenherds. Moreover, the underlying assumption about normality of a linear model is notfulfilled in data on threshold characteristics. Disease traits are basically coded as 1 or0 depending on whether a disease was detected or not. Prediction and estimationprocedures based on normality are approximative and may yield poor results (Gröhn etal., 1986). Difficulties in analysing all-or-none traits and the highly multi-factorialnature of diseases may very well account for controversial results found by differentauthors.4. A biological explanation for the occurrence of negative genetic correlationsIn 1954, Lerner discussed what he called ‘genetic homeostasis’, the idea that evolutionby means of natural selection has developed genotypes that are highly adapted to theirenvironment, since the most adapted phenotypes in a population will predominate.Heterozygosity, stabilised selection and negative genetic correlations between traitswill result in intermediate optima for many characteristics in order to maintain thishomeostasis. All traits will maintain additive variance which will act as a bufferingeffect to a wide range of environments, allowing species to genetically change rapidly ifthe conditions alter. In 1963, Rendel suggested that genetic correlations may be explained from asituation in which two characters share resources for their development. An increase inresources will result in an increase in both traits and a consequent positive correlation;a limited resource situation, resulting in competition for resources among traits, willresult in a negative correlation. Rendel (1963) suggests that selection in order toincrease a certain trait may act on both the total amount of resources available andtheir distribution. These ideas were worked out further by Goddard and Beilharz (1977) who relatedtotal amount of resources available to an animal to fitness in the ‘Resource AllocationTheory’: fitness is a trait composed of several components, such as ‘number of parities’and ‘average litter size’, which they suggest multiply to give fitness. The resourcesconsumed by these and other processes (maintenance, (re)production, movement,reaction to pathogens and stressors, etc.) they suggest, add to give the total amount ofresources consumed (e.g., food intake, body tissue, etc.), since resources consumed byone process can not be allocated to another process. In a limited resource situation,fitness will decrease if one of its components increases in combination with anincreased allocation of resources to this trait. In this situation, fitness will reach alimit with optimal intermediate values for its components (Beilharz et al., 1993). It issuggested that the process of domestication has increased total amount of resourcesavailable to an animal since some resource consuming traits are no longer required, e.g.,
    • 2. Undesirable side effects of selection for high production efficiency 23searching for food, fighting against predators, etc. However, Beilharz et al. (1993)suggest that we must expect domesticated animals to have become again limited by theenvironment in many situations. Residual food intake (RFI), which is the part of foodintake that is not accounted for by maintenance and (re)production, is suggested as apossible tool to quantify the total amount of resources available to an animal for otherprocesses than maintenance and (re)production, i.e., ‘movement’, ‘reaction to pathogens’,‘reaction to stress’, etc. in different environments and different metabolical stages oflife (Luiting et al., 1995; Luiting et al., 1997). Artificial selection for a particular trait may lead to the situation in whichresources are used to the maximum, i.e., no buffer is left to respond adequately tounexpected stresses and challenges. Preferential allocation of resources may occurbecause the animal may be ‘genetically pre-programmed’ to allocate a disproportionallylarge amount of resources to the trait selected for, leaving the animal lacking in abilityto respond to other demands. If genetic changes are too radical or sought too rapidly,the population may lack the time required to adapt to the changes imposed on it byselection and the homeostatic balance of the animal is at risk (Dunnington, 1990). Thegreater the loss of balance, the more negative genetic correlations between productiontraits and fitness traits will become in order to counteract this loss. Welfare of an animal is reflected by the success of its attempts to cope with itsenvironment (Broom, 1993), and depends on its physiological ability to respond properlyin order to maintain or re-establish its homeostatic state or balance (Siegel, 1995). Anystimulus that challenges homeostasis can be viewed as a stressor, and changes inbiological function occur as the animal attempts to respond to the stressor. Alterationsin physiological systems to maintain homeostatic balance will divert resources from thebiological functions that occur prior to the stress, which may lead to the developmentof pathology, e.g., increased susceptibility to disease, impaired reproduction orinefficient metabolism, and thus impaired animal welfare (Moberg, 1985, 1987; Newman,1994).5. Discussion and conclusionsThe examples presented in this review show that, apart from a highly favourableincrease in production, present-day selection for high production efficiency in livestockspecies in many cases has been accompanied by undesirable side effects for severalphysiological, immunological and reproduction traits and consequently for animalwelfare. Although the continuous increase in production levels suggests a continuouspositive selection response for the traits under selection in commercial livestockproduction, and desirable correlated responses to selection are likely to exist as well, itis expected that with ongoing selection for high production efficiency only, theoccurrence and magnitude of undesirable side effects will increase. The most striking examples of undesirable correlated responses are found inbroiler chickens with an increasing incidence of heart failure syndrome and legproblems. In poultry breeding programs selection has been almost for one trait only,i.e., body weight at a certain age, with a high selection intensity and a short generationinterval. In species like cattle and pigs, where the results are obviously more
    • 24 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommerscontroversial, selection has been less intensive, for more traits and during fewergenerations. Moreover, especially in dairy cattle, undesirable side effects may becamouflaged because of the multi-factorial nature of most problems, which does not,however, mean that genetic relationships between production traits and undesirabletraits are absent. Since many literature references do show the presence ofundesirable side-effects of selection in dairy cattle, it may be concluded that selectionhas indeed made also these animals more sensitive to metabolic, reproduction andhealth problems. Future application of DNA-technology and modern reproduction technologies mayincrease production faster than at present. This will, however, also increase the speedof expression of detrimental antagonistic relationships with selection for highproduction efficiency. In this perspective, the undesirable effects found in broilersmay be considered as a forerunner for similar problems to be expected in other speciesunder selection. It is important to avoid this for several reasons: 1) there is increasingconsciousness among people of the intensive nature of animal production systems andsocial resistance against some production systems will increase if animal health andwelfare become more at risk, 2) veterinary costs and costs of replacing animals willfurther increase, 3) if breeding programs should be altered it may take five to tenyears before genetic trends in commercial livestock are really changed. In an evolutionary perspective, in nature those animals with high fitness have ahigher contribution of alleles to the population in future generations than animals withlow fitness. Consequently, alleles related to high fitness (i.e., health, reproduction,longevity, etc.) replace alleles related to low fitness. According to the ResourceAllocation Theory of Beilharz et al. (1993), when (internal and/or external) resourcesare limited, a compromise has to be found how to partition available resources amongtraits. Assuming that in nature alleles related to highest fitness predominate, theoptimal partitioning of resources in nature is accomplished by allocating intermediateproportions to fitness traits which will maximise overall fitness. With artificial selection, however, the ‘optimal situation’ is being redefinedtowards high production. Moreover, fitness does not necessarily have to be as definedin nature, e.g., long reproductive life may not be necessary, but animals have to be(reasonably) healthy and reproductive. The theory implies that once a breeding goal hasbeen defined, there is an optimum to what can be accomplished in a given, resourcelimited, environment. Increasing production by selection beyond this optimum will becompromised because the environment is not able to support the essential increase inresources required, resulting in a deviation from the optimum (Beilharz, 1998). When apopulation is genetically driven towards high production, and thus allocating a higherproportion of resources to this trait, less resources will be left to respond adequatelyto other demands, like coping with (unexpected) stressors; i.e., the buffer capacity isaffected. In this situation it is most likely that those traits not defined in the breedinggoal will be the first ones from which resources will be diverted towards increasedproduction. The theory implies that a set of alleles in a given environment, maximallysupporting the breeding goal by optimally allocating available resources to the traitsincluded, will not maximally support the breeding goal in a different environment. High
    • 2. Undesirable side effects of selection for high production efficiency 25producing Western dairy breeds are not able to fully express their potentially highproduction level while maintaining the level of health and reproduction when they haveto deal with heat stress, diseases and limited nutrient supply in the tropics, as is wellrecognised (e.g., Syrstad, 1989; Menendez Buxadera and Dempfle, 1997). Consequenceswill be less pronounced for differences only in management systems, food composition,etc. When the Resource Allocation Theory (Beilharz et al., 1993) holds, and futureresearch into the quantitative aspects of resource allocation is desired, negative sideeffects of selection may be predicted and thus prevented. Modification of theenvironment to increase the amount of resources available to an animal, e.g., byincreasing the energy amount of feedstuffs or by reducing environmental stress (e.g.,Specific Pathogen Free environments), may prevent negative side effects of selectionor even allow for improved output levels. However, possibilities to change theenvironment are limited and costly, and the population may become more dependent onthe specific environment. Furthermore, selection for increased food intake capacitymay increase the resource situation. Efficiency parameters have to be handled withcare: although increasing net efficiency of a specific trait may improve the resourcesituation, increasing food efficiency for production may be (partly) at the cost ofresources left to respond to stressors (e.g., maintenance requirements) and may thushave the opposite effect. A more fundamental solution is to redefine the breeding goal into a broaderperspective. It means breeding animals with a long economical (re)productive life at aproduction level that is economical (i.e., production in relation to veterinary costs, etc.)without giving any signs of disturbed welfare. Those traits will have to be closelydefined. Livestock breeders will have to be satisfied with a slower increase in(re)production. Moreover, animals will have to be selected preferably within specific environments(e.g., regions, management systems) for selection to be most effective, since differentenvironments have different optima. Mathematically accounting for differences inenvironmental conditions may only result in a further deviation from the optimum whenanimals end up in an environment that can not support the required amount of resourcesto fully express the genetic potential. Finally, when the optimum is reached, i.e., the breeding goal is maximally supported,production can be further increased only when the resource situation (internal and/orexternal) is improved or the breeding goal is redefined. Although the economic importance of genetic changes in production traits is clear,relatively little research has been directed to the biological aspects of selection.Animal breeding scientists have concentrated mainly on the technical aspects ofbreeding value and genetic parameter estimation. Without knowledge about theunderlying physiological processes on which genetic selection acts, cumulative andpermanent genetic improvement through selection is essentially a black box technique.Speeding up genetic increase, e.g., with application of modern reproduction techniquesand DNA-technology, in a biological system that is not well understood is very likely tolead to unfavourable and improperly understood side effects, if not to disorders(Luiting, 1993). Knowledge of biological backgrounds, including typing of genes and
    • 26 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommersidentifying their specific role within physiology, will offer the opportunity tounderstand, anticipate and prevent negative side effects of selection.
    • 2. Undesirable side effects of selection for high production efficiency 27References• Anthony, N.B., Dunnington, E.A., Siegel, P.B., 1989. Egg production and egg composition of parental lines and F1 and F2 crosses of White Rock chickens selected for 56-day body weight. Poultry Sci. 68:27-36.• Anthony, N.B., Emmerson, D.A., Nestor, K.E., Bacon, W.L., Siegel, P.B., Dunnington, E.A, 1991. Comparison of growth curves of weight selected populations of turkeys, quail and chickens. Poultry Sci. 70:13-19.• Barbato, G.F., Siegel, P.B., Cherry, J.A., Nir, I., 1984. Selection for body weight at eight weeks of age. 17. Overfeeding. Poultry Sci. 63:11-18.• Barbut, S., 1997. Occurrence of pale soft exudative meat in mature turkey hens. Brit. Poultry Sci. 38:74-77.• Barnes, M.A., Kazmer, G.W., Akers, R.M., Pearson, R.E., 1985. Influence of selection for milk yield on endogenous hormones and metabolites in Holstein heifers and cows. J. Anim. Sci. 60:271-284.• Barria, N. and Bradford, G.E., 1981. Long-term selection for rapid gain in mice. I Genetic analysis at the limit of response. J. Anim. Sci. 52:729-738.• Beilharz, R.G., 1998. The problem of genetic improvement when environments are limiting. Proc. 6th World Congr. Gen. Appl. Livest. Prod., Armidale, Australia.• Beilharz, R.G., Luxford, B.G., Wilkinson, J.L., 1993. Quantitative genetics and evolution: Is our understanding of genetics sufficient to explain evolution? J. Anim. Breed. Gen. 110:161-170.• Berger, P.J., Shanks, R.D., Freeman, A.E., Laben, R.C., 1981. Genetic aspects of milk yield and reproductive performance. J. Dairy Sci. 64:114-122.• Bonczek, R.R., Richardson, D.O., Moore, E.D., Miller, R.H., Owen, J.R., Dowlen, H.H., Bell, B.R., 1992. Correlated responses in reproduction accompanying selection for milk yield in Jerseys. J. Dairy Sci. 75:1154-1160.• Bonczek, R.R., Young, C.W., Wheaton, J.E., Miller, K.P., 1988. Responses of somatotropin, insulin, prolactin, and thyroxine to selection for milk yield in Holsteins. J. Dairy Sci. 71:2470- 2479.• Broom, D.M., 1993. Assessing the welfare of modified or treated animals. Livest. Prod. Sci. 36:39-54.• Bunch, K.J., Heneghan, D.J.S., Hibbitt, K.G., Rowlands, G.J., 1984. Genetic influences on clinical mastitis and its relationship with milk yield, season, and stage of lactation. Livest. Prod. Sci. 11:91-104.• Burkhart, C.A., Cherry, J.A., Van Krey, H.P., Siegel, P.B., 1983. Genetic selection for growth rate alters hypothalamic satiety mechanisms in chickens. Behav. Genet. 13:295-300.• Butler, W.R., Everett, R.W., Coppock, C.E., 1981. The relationships between energy balance, milk production and ovulation in postpartum Holstein cows. J. Anim. Sci. 53:742-748.• Butler, W.R. and Smith, R.D., 1989. Interrelationships between energy balance and postpartum reproductive function in dairy cattle. J. Dairy Sci. 72:767-783.• Bünger, L., Schüler, L., Renne, U., 1984. Zur Selektion auf Wachstum nach vorheriger Fruchtbarkeitsselektion bei Labormäusen. Archiv für Tierzucht 27:183-191.• Cahaner, A. and Siegel, P.B., 1986. Evaluation of industry breeding programmes for meat-type chickens and turkeys. Proc. 3rd World Congr. Genet. Appl. Livest. Prod. (X):337-346.• Calabotta, D.F., Cherry, J.A., Siegel, P.B., Jones, D.E., 1985. Lipogenesis and lipolysis in fed and fasted chicks from high and low body weight lines. Poultry Sci. 64:700-704.
    • 28 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers• Canfield, R.W. and Butler, W.R., 1990. Energy balance and pulsatile LH secretion in early postpartum dairy catle. Dom. Anim. Endocr. 7:323-330.• Canfield, R.W. and Butler, W.R., 1991. Energy balance, first ovulation and the effects of naloxone on LH secretion in early postpartum dairy cows. J. Anim. Sci. 69:740-746.• Canon, J., Gruand, J., Gutierrez, J.P., Ollivier, L., 1992. A repeat-sire mating selection experiment on lean tissue growth in the pig: genetic trends of the traits selected upon. Genet. Sel. Evol. 24:449-462.• Chambers, J.R., Gavora, J.S., Fortin, A., 1981. Genetic changes in meat-type chickens in the last twenty years. Can. J. Anim. Sci. 61:555-563.• Cheng, K.M. and Goodman, B.L., 1976. The influence of divergent growth selection on semen traits, fertility and hatchability. Poultry Sci. 55:457-459.• Cherry, J.A. and Siegel, P.B., 1978. Selection for body weight at eight weeks of age. 15. Feed passage and intestinal size of normal and dwarf chicks. Poultry Sci. 57:336-340.• Denbow, D.M., Van Krey, H.P., Siegel, P.B., 1986. Selection for growth alters the feeding response to injections of biogenic amines. Pharmacol. Biochem. Behav. 24:39-42.• Ducker, M.J. and Morant, S.V., 1984. Observations on the relationships between the nutrition, milk yield, live weight and reproductive performance of dairy cows. Anim. Prod. 38:9-14.• Dunnington, E.A., 1990. Selection and homeostasis. Proc. 4th World Congr. Gen. Appl. Livest. Prod., Edinburgh, UK (XVI):5-12.• Dunnington, E.A. and Siegel, P.B., 1995. Enzyme activity and organ development in newly hatched chicks selected for high or low eight-week body weight. Poultry Sci. 74:761-770.• Dunnington, E.A. and Siegel, P.B., 1996. Long-term divergent selection for eight-week body weight in White Plymouth Rock chickens. Poultry Sci. 75:1168-1179.• Edens, F.W., Van Krey, H.P., Siegel, P.B., 1973. Selection for body weight at eight weeks of age. 9. Chickens spermatozoal respiration. Poultry Sci. 52:977-981.• Eicker, S.W., Gröhn, Y.T., Hertl, J.A., 1996. The association between cumulative milk yield, days open, and days to first breeding in New York Holstein cows. J. Dairy Sci. 79:235-241.• Eisen, E.J., 1980. Conclusions from long-term selection experiments with mice. Z. Tierzüchtg. Züchtgsbiol. 97:305-319.• Essén-Gustavsson, B. and Lindholm, A., 1984. Fiber types and metabolic characteristics in muscles of wild boars, normal and halothane sensitive Swedish Landrace pigs. Comp. Biochem. Physiol. 78A:67-71.• Falconer, D.S. and Mackay, T.F.C., 1996. Introduction to quantitative genetics. 4th ed., Longman Group Ltd, UK.• Fonseca, F.A., Britt, J.H., McDaniel, B.T., Wilk, J.C., Rakes, A.H., 1983. Reproductive traits of Holstein and Jerseys. Effects of age, milk yield, and clinical abnormalities on involution of cervix and uterus, ovulation, estrus cycles, detection of estrus, conception rate, and days open. J. Dairy Sci. 66:1128-1147.• Freeman, A.E., 1986. Genetic control of reproduction and lactation in dairy cattle. Proc. 3rd World Congr. Gen. Appl. Livest. Prod. 11:3-13.• Fredeen, H.T., 1984. Selection limits: have they been reached with pigs? Can. J. Anim. Sci. 64:223-234.• Gallo, L., Carnier, P., Cassandro, M., Mantovani, R., Bailoni, L., Contiero, B., Bittante, G., 1996. Change in body condition score of Holstein cows as affected by parity and mature equivalent milk yield. J. Dairy Sci. 79:1009-1015.
    • 2. Undesirable side effects of selection for high production efficiency 29• Goddard, M.E. and Beilharz, R.G., 1977. Natural selection and animal breeding. Proc. 3rd Int. Congr. S.A.B.R.A.O., Animal Breeding Papers 4.19 to 4.21.• Gröhn, Y.T., Saloniemi, H.S., Syväjärvi, J., 1986. An epidemiological and genetic study on registered diseases in Finnish Ayrshire cattle. III. Metabolic diseases. Acta Vet. Scand. 27:209-222.• Grommers, F.J., 1987. Limits of productivity in dairy cattle: heath aspects. 38th Annual Meeting of the EAAP, Lisbon, Portugal.• Hansel, W. and Godke, R.A., 1992. Future prospectives on animal biotechnology. Anim. Biotechn. 3:111-137.• Hansen, L.B., Freeman, A.E., Berger, P.J., 1983. Yield and fertility relationships in dairy cattle. J. Dairy Sci. 66:293-305.• Harrison, R.O., Ford, S.P., Young, J.W., Conley, A.J., Freeman, A.E., 1990. Increased milk production versus reproductive and energy status of high producing dairy cows. J. Dairy Sci. 73:2749-2758.• Harrison, R.O., Young, J.W., Freeman, A.E., Ford, S.P., 1989. Effects of lactational level on reactivation of ovarian function, and interval from parturition to first visual oestrus and conception in high-producing Holstein cows. Anim. Prod. 49:23-28.• Havenstein, G.B., Ferket, P.R., Scheideler, S.E., Larson, B.T., 1994a. Growth, livability, and feed conversion of 1957 vs 1991 broilers when fed ‘typical’ 1957 and 1991 broiler diets. Poultry Sci. 73:1785-1794.• Havenstein, G.B., Ferket, P.R., Scheideler, S.E., Rives, D.V., 1994b. Carcass composition and yield in 1991 vs 1957 broilers when fed ‘typical’ 1957 and 1991 broiler diets. Poultry Sci. 73:1795-1804.• Henckel, P., 1992. Properties of muscle fibre types as a source of variation in meat quality. Proc. 19th World’s Poultry Congr., Amsterdam, The Netherlands. Suppl.:87-92.• Henckel, P., Jørgensen, P.F., Jensen, P., 1992. Glycogen content, buffering capacity and resting pH in live muscles of pigs of different halothane genotypes (a pilot project). Meat Sci. 32:131- 138.• Hoekstra, J., Van der Lugt, A.W., Van der Werf, J.H.J., Ouweltjes, W., 1994. Genetic and phenotypic parameters for milk production and fertility traits in upgraded dairy cattle. Livest. Prod. Sci. 40:225-232.• Huang, S.Y., Tsou, H.L., Kan, M.T., Lin, W.K., Chi, C.S., 1995. Genetic study on leg weakness and its relationship with economic traits in central tested boars in subtropical area. Livest. Prod. Sci. 44:53-59.• Hunton, P., 1984. Selection limits: have they been reached in the poultry industry? Can. J. Anim. Sci. 64:217-221.• Hutchens, L.K., Hintz, R.L., Johnson, R.K., 1981. Genetic and phenotypic relationships between pubertal and growth characteristics of gilts. J. Anim. Sci. 53:946-951.• IKC, 1996. Dutch piglet price schemes from 1973 to 1995. Informatie en kennis centrum Landbouw, Ede, The Netherlands.• Jaap, R.G. and Muir, F.V., 1968. Erratic oviposition and egg defects in broiler-type pullets. Poultry Sci. 47:417-423.• Karlström, K., 1995. Capillary supply, fibre type composition and enzymatic profile of equine, bovine and porcine locomotor and nonlocomotor muscles. PhD thesis Swedish University of Agricultural Sciences, Uppsala, Sweden.
    • 30 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers• Kennedy, B.W., 1984. Selection limits: have they been reached with the dairy cow? Can. J. Anim. Sci. 64:207-215.• Kerr, J.C. and Cameron, N.D., 1995. Reproductive performance of pigs selected for components of efficient lean growth. Anim. Sci. 60:281-290.• Kerr, J.C. and Cameron, N.D., 1996. Genetic and phenotypic relationships between performance test and reproduction traits in Large White pigs. Anim. Sci. 62:531-540.• Kirkwood, R.N. and Aherne, F.X., 1985. Energy intake, body composition and reproductive performance of gilts. J. Anim. Sci. 60:1518-1529.• Leenstra, F., 1993. Selection for disease resistance in poultry: general or specific? Arch. Geflügelk. 57:69-73.• Legates, J.E. and Myers, R.M., 1988. Measuring genetic change in a dairy herd using a control population. J. Dairy Sci. 71:1025-1033.• LEI-DLO, 1996. Landbouwcijfers 1996, Table 83f, Den Haag, The Netherlands.• Lepore, P.D., Siegel, P.B., King, K.W., 1963. Proximate and amino acid composition of eggs and chicks from growth-selected lines of White Rocks. Life Sciences 2:584.• Lerner, I.M., 1954. Genetic Homeostasis. John Wiley & Sons, Inc., New York.• Liu, G., Dunnington, E.A., Siegel, P.B., 1995. Correlated responses to long-term divergent selection for eight-week body weight in chickens: growth, sexual maturity, and egg production. Poultry Sci. 74:1259-1268.• Luiting, P., 1990. Genetic variation of energy partitioning in laying hens: causes of variation in residual feed consumption. World’s Poult. Sci. J. 46:133-152.• Luiting, P., 1993. The biological nature of genetic variation in net feed efficiency. 44th Annual Meeting of the EAAP, Aarhus, Denmark.• Luiting, P., Bakken, M., Rauw, W.M., De Veer, C.J.M., 1995. Behavioural differences in a long- term selection experiment for litter size in mice. 47th Annual Meeting of the EAAP, Prague, Czech Republic.• Luiting, P., Vangen, O., Rauw, W.M., Knap, P.W., Beilharz, R.G., 1997. Physiological consequences of selection for growth. 48th Annual Meeting of the EAAP, Vienna, Austria.• Lundeheim, N., 1987. Genetic analysis of osteochondrosis and leg weakness in the Swedish pig progeny testing scheme. Acta Agric. Scand. 37:159-173.• Lyons, D.T., Freeman, A.E., Kuck, A.L., 1991. Genetics of health traits in Holstein cattle. J. Dairy Sci. 74:1092-1100.• Maloney, M.A., Gilbreath, J.C., Tierce, J.F., Morrison, R.D., 1967. Divergent selection for twelve-week body weight in the domestic fowl. Poultry Sci. 46:1116-1127.• Marini, P.J. and Goodman, B.L., 1969. Semen characteristics as influenced by selection for divergent growth rate in chickens. Poultry Sci. 48:859-865.• Marks, H.L., 1979. Growth rate and feed intake of selected and nonselected broilers. Growth 43:80-90.• McCarthy, J.C. and Siegel, P.B., 1983. A review of genetical and physiological effects of selection in meat-type poultry. Anim. Breed. Abstr. 51:87-94.• McGowan, M.R., Veerkamp, R.F., Anderson, L., 1996. Effects of genotype and feeding system on the reproductive performance of dairy cattle. Livest. Prod. Sci. 46:33-40.• Menendez Buxadera, A., Dempfle, L., 1997. Genetic and environmental factors affecting some reproductive traits of Holstein cows in Cuba. Genet. Sel. Evol. 29:469-482.
    • 2. Undesirable side effects of selection for high production efficiency 31• Miller, L.L., Siegel, P.B., Dunnington, E.A., 1992. Inheritance of antibody response to sheep erythrocytes in lines of chickens divergently selected for 56-day body weight and their crosses. Poultry Sci. 71:47-52.• Moberg, G.P., 1985. Biological response to stress: key to assessment of animal well-being? In: Animal Stress. Moberg, G.P., ed. American physiological society, Maryland. p. 27-49.• Moberg, G.P., 1987. A model for assessing the impact of behavioral stress on domestic animals. J. Anim. Sci. 65:1228-1235.• Nebel, R.L. and McGilliard, M.L., 1993. Interactions of high milk yield and reproductive performance in dairy cows. J. Dairy Sci. 76:3257-3268.• Nestor, K.E., 1977. The influence of a genetic change in egg production, body weight, fertility or response to cold stress on semen yield in the turkey. Poultry Sci. 56:421-425.• Nestor, K.E., Noble, D.O., Zhu, J., Moritsu, Y., 1996a. Direct and correlated responses to long- term selection for increased body weight and egg production in turkeys. Poultry Sci. 75:1180- 1191.• Nestor, K.E., Saif, Y.M., Zhu, J., Noble, D.O., 1996b. Influence of growth selection in turkeys on resistance to Pasteurella multocida. Poultry Sci. 75:1161-1163.• Newman, S., 1994. Quantitative- and molecular-genetic effects on animal well-being: adaptive mechanisms. J. Anim. Sci. 72:1641-1653.• Nir, I., Nitzan, Z., Dror, Y., Shapira, N., 1978. Influence of overfeeding on growth, obesity and intestinal tract in young chicks of light and heavy breeds. Br. J. Nutr. 39:27-35.• NML/Norske Meierier, 1996. Annual report 1996, Norway.• Norsvin, 1996. Annual reports 1980-1996. Norwegian Pig Breeding Organisation, Hamar, Norway.• NRS, 1996. Annual statistics 1996, Royal Dutch Cattle Syndicate, Arnhem, The Netherlands.• Owens, C.A., Siegel, P.B., Van Krey, H.P., 1971. Selection for body weight at 8 weeks of age. 8. Growth and metabolism in two light environments. Poultry Sci. 50:548-553.• Pirchner, F., 1983. Population genetics in animal breeding. 2nd ed., Plenum Press, New York.• Pryce, J.E., Veerkamp, R.F., Thompson, R., Hill, W.G., Simm, G., 1997. Genetic aspects of common health disorders and measures of fertility in Holstein Friesian dairy cattle. Anim. Sci. 65:353-360.• Pösö, J. and Mäntysaari, E.A., 1996. Relationships between clinical mastitis, somatic cell score, and production for the first three lactations of Finnish Ayrshire. J. Dairy Sci. 79:1284-1291.• Qureshi, M.A. and Havenstein, G.B., 1994. A comparison of the immune performance of a 1991 commercial broiler with a 1957 randombred strain when fed ‘typical’ 1957 and 1991 broiler diets. Poultry Sci. 73:1805-1812.• Raheja, K.L., Burnside, E.B., Schaeffer, L.R., 1989. Relationship between fertility and production in Holstein dairy cattle in different lactations. J. Dairy Sci. 72:2670-2678.• Rahelic, S. and Puac, S., 1980. Fibre types in Longissimus Dorsi from wild and highly selected pig breeds. Meat Sci. 5:439-450.• Reddy, R.P., 1996. Symposium: the effects of long-term selection on growth of poultry - Introduction. Poultry Sci. 75:1164-1167.• Reddy, P.R.K. and Siegel, P.B., 1977. Chromosomal abnormalities in chickens selected for high and low body weight. J. Heredity 68:253-256.• Rendel, J.M., 1963. Correlation between the number of scutellar and abdominal bristles in drosophila melanogaster. Genetics 48:391-408.
    • 32 W.M. Rauw, E. Kanis, E. N. Noordhuizen-Stassen and F.J. Grommers• Rydhmer, L., Eliasson-Selling, L., Johansson, K., Stern, S., Andersson, K., 1994. A genetic study of estrus symptoms at puberty and their relationship to growth and leanness in gilts. J. Anim. Sci. 72:1964-1970.• Rydhmer, L., Johansson, K., Stern, S., Eliasson-Selling, L., 1992. A genetic study of pubertal age, litter traits, weight loss during lactation and relations to growth and leanness in gilts. Acta Agric. Scand., Sect. A, Anim. Sci. 42:211-219.• Rydhmer, L., Lundeheim, N., Johansson, K., 1995. Genetic parameters for reproduction traits in sows and relations to performance-test measurements. J. Anim. Breed. Genet. 112:33-42.• Sather, A.P., 1987. A note on the changes in leg weakness in pigs after being transferred from confinement housing to pasture lots. Anim. Prod. 44:450-453.• Scheele, C.W., 1996. Ascites in chickens. Oxygen consumption and requirement related to its occurrence. PhD Thesis Wageningen Agricultural University, The Netherlands.• Scheele, C.W., Decuypere, E., Vereijken, P.F.G., Schreurs, F.J.G., 1992. Ascites in broilers. 2. Disturbances in the hormonal regulation of metabolic rate and fat metabolism. Poultry Sci. 71:1971-1984.• Schneeberger, M. and Hagger, C., 1986. Relationship of fertility parameters with lactation yield in cows of various crossbreeding levels. 3rd World Congr. Gen. Appl. Livest. Prod., Lincoln, Nebraska, USA, p. 107-112.• Senatore, E.M., Butler, W.R., Oltenacu, P.A., 1996. Relationships between energy balance and post-partum ovarian activity and fertility in first lactation dairy cows. Anim. Sci. 62:17-23.• Seykora, A.J. and McDaniel, B.T., 1983. Heritabilities and correlations of lactation yields and fertility for Holsteins. J. Dairy Sci. 66:1486-1493.• Shanks, R.D., Freeman, A.E., Berger, P.J., Kelley, D.H., 1978. Effect of selection for milk production on reproductive and general health of the dairy cow. J. Dairy Sci. 61:1765-1772.• Siegel, H.S., 1995. Stress, strains and resistance. Br. Poultry. Sci. 36:3-22.• Siegel, P.B., 1963. Selection for body weight at eight weeks of age. 2. Correlated responses of feathering, body weights and reproductive characteristics. Poultry Sci. 42:896-905.• Siegel, P.B. and Dunnington, E.A., 1985. Reproductive complications associated with selection for broiler growth. In: Poultry genetics and breeding. W.G. Hill, J.H. Manson and D. Hewitt, ed. Longman Group, Harlow, United Kingdom. p. 59-72.• Simianer, H., Solbu, H., Schaeffer, L.R., 1991. Estimated genetic correlations between disease and yield traits in dairy cattle. J. Dairy Sci. 74:4358-4365.• Sinsigalli, N.A., McMurtry, J.P., Cherry, J.A., Siegel, P.B., 1987. Glucose tolerance, plasma insulin and immunoreactive glucagon in chickens selected for high and low body weight. J. Nutr. 117:941-947.• Soller, M., 1994. Marker assisted selection - an overview. Anim. Biotechn. 5:193-207.• Syrstad, O., 1989. Dairy cattle cross-breeding in the tropics: performance of secondary cross-bred populations. Livest. Prod. Sci. 23:97-106.• Syväjärvi, J., Saloniemi, H.S., Gröhn, Y.T., 1986. An epidemiological and genetic study on registered diseases in Finnish Ayrshire cattle. IV. Clinical mastitis. Acta Vet. Scand. 27:223- 234.• Ten Napel, J., 1996. Genetic aspects of intervals from weaning to estrus in swine. PhD thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Ten Napel, J. and Johnson, R., 1997. Genetic relationships among production traits and rebreeding performance. J. Anim. Sci 75:51-60.
    • 2. Undesirable side effects of selection for high production efficiency 33• Ten Napel, J., Kemp, B., Luiting, P., De Vries, A.G., 1995. A biological approach to examine genetic variation in weaning-to-oestrus interval in first-litter sows: a review. Livest. Prod. Sci. 41:81-93.• Thompson, J.R., 1984. Genetic interrelationships of parturition problems and production. J. Dairy Sci. 67:628-635.• Uribe, H.A., Kennedy, B.W., Martin, S.W., Kelton, D.F., 1995. Genetic parameters for common health disorders of Holstein cows. J. Dairy Sci. 78:421-430.• USDA, 1991. Annual Summaries, Federal Milk Order Market Statistics, AMS.• USDA, 1995a. Poultry yearbook, Table 71 ‘young chicken: average liveweight at slaughter under Federal inspection 1960-95, Poultry Slaughter National Agricultural Statistics Service, United States Department of Agriculture.• USDA, 1995b. Poultry yearbook, Table 138 ‘turkey: average liveweight at slaughter under Federal inspection 1960-95, Poultry Slaughter National Agricultural Statistics Service, United States Department of Agriculture.• Van Arendonk, J.A.M., Hovenier, R., De Boer, W., 1989. Phenotypic and genetic association between fertility and production in dairy cows. Livest. Prod. Sci. 21:1-12.• Van Middelkoop, J.H. and Siegel, P.B., 1976. Classification of abnormal chicken eggs. Poultry Sci. 55:1563-1566.• Veerkamp, R.F., Simm, G., Oldham, J.D., 1994. Effects of interaction between genotype and feeding system on milk production, feed intake, efficiency and body tissue mobilization in dairy cows. Livest. Prod. Sci. 39:229-241.• Villa-Godoy, A., Hughes, T.L., Emery, R.S., Chapin, L.T., Fogwell, R.L., 1988. Association between energy balance and luteal function in lactating dairy cows. J. Dairy Sci. 71:1063-1072.• Wautlet, R.G., Hansen, L.B., Young, C.W., Chester-Jones, H., Marx, G.D., 1990. Calving disorders of primiparous Holsteins from designed selection studies. J. Dairy Sci. 73:2555- 2562.• Webb, A.J., Russell, W.S., Sales, D.I., 1983. Genetics of leg weakness in performance-tested boars. Anim. Prod. 36:117-130.• Woltmann, M.D., Clutter, A.C., Buchanan, D.S., 1995. Effect of divergent selection for postweaning average daily gain on front-end soundness of market-weight pigs. J. Anim. Sci. 73:1940-1947.• Whittemore, C.T., 1994. Causes and consequences of change in the mature size of the domestic pig. Outlook on Agr. 23:55-59.
    • 3 FOOD RESOURCE ALLOCATION PATTERNS IN NON-REPRODUCTIVE MALES AND FEMALES. I. TRENDS IN BODY WEIGHT AND FOOD INTAKE AGAINST TIME W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. KnapAbstract Differences in the metabolic resource situation between non-reproductivemale and female mice of a line selected for high litter size at birth (average of 22 bornper litter) and a non-selected control line (average of 10 born per litter) wereinvestigated in two replicates. Brody curves were fitted to individual data on bodyweight against age and linear regression lines were fitted to individual data oncumulative food intake against age. Mature body weight and mature daily food intakewere higher in selected mice than in control mice and higher in males than in females.Selected males matured faster than selected females and control mice. In general,differences in growth and food intake curves between species or lines can mostly beexplained by differences in mature size. Therefore, parameters were subsequentlyscaled by individual estimates of mature body weight. Differences that remain afterscaling are a consequence of what have been called specific genetic factors. Scaledmature food intake was higher in selected mice than in control mice and higher infemales than in males. Scaled maturation rate was higher in selected mice than incontrol mice and higher in selected males than in selected females. This shows that inthe present study, specific genetic factors have been detected for both body weightand food intake, which suggests that selection for increased litter size hasdisproportionally changed the resource allocation pattern.Keywords: genetic size scaling, litter size, resource allocation, selectionBased on Animal Science 71, W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knap,Differences in food resource allocation in a long-term selection experiment for litter size inmice. I. Developmental trends in body weight and food intake against time, 31-38, ©2000 BritishSociety of Animal Science, with permission from British Society of Animal Science.
    • 36 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knap1. IntroductionSelection for high body weight leads usually to a higher litter size (Taylor and Murray,1987). In many data sets this correlated response disappears after correction formature body weight (A), which means that the higher litter size is a scale effect only(Taylor and Murray, 1987). The reverse situation has also been observed: selection forhigh litter size results in higher body weights and food intakes (e.g., De la Fuente andSan Primitivo, 1985; Narayan and Rawat, 1986), which suggests that the correlatedresponse in body weight to selection for high litter size may also be a scale effect only. The aim of the present study is to investigate the metabolic resource situation innon-reproductive males and females in a long-term selection experiment for litter sizein mice. The present paper aims at visualising and detecting differences in the resourcesituation; the accompanying paper (Chapter 4) aims at quantifying and explaining thedifferences.2. TheoryWhen comparing growth curves of different species, many of the sigmoid shapedcurves look very similar but at a different scale. It appears that animals growaccording to a general standard life programme, which determines the animal’s designfrom conception to maturity (Ogink, 1993). In 1965, Taylor developed the concept ofone uniform time-scaling unit for maturation intervals. Specific ontogenetic events, likeweaning age and sexual maturity, and their intervals, like gestation length or lifespan,correspond to similar fractions of mature body weight among species. Taylor (1980)generalised this concept with the genetic size-scaling theory, a set of scaling ruleswhich relate all traits associated with growth and metabolism to a genetic size factoras a general procedure to describe the similarities in all aspects of the growth processof different genotypes (Ogink, 1993). The inherent genotype-specific genetic sizefactor, estimated as adult body weight (A), operates throughout growth and is mostclearly expressed in the fully grown adult (Taylor, 1985). Two genetic size-scaling rules were formulated: (1) all age and time variables areproportional to A0.27, where age is measured in days from an origin near (i.e., 3.5 daysafter) conception (Taylor (1965 and 1980) adjusted age in days from conceptionempirically to an age origin 3.5 days after conception, which corresponds approximatelyto the average time taken by fertilised eggs to reach the uterus) and (2) all cumulatedgrowth variables, such as body weight and cumulative food intake, are proportional to A.As a consequence, rate expressed traits, such as growth rate and daily food intake, areproportional to A0.73. The size-scaling rules also imply that efficiency parameters, i.e.,ratios between cumulated or rate-expressed input and output variables, areindependent of size. Consequently, efficiency of energy utilisation for production issimilar in animals of different sizes (Taylor, 1980; Ogink, 1993). Two major outcomes of the scaling rules are ‘metabolic age (θ)’, i.e., age scaled by 0.27A , and ‘degree of maturity in body weight (u)’, i.e., body weight scaled by A (Taylor,1980). Observed phenotypic variation of a trait (σ2P) may not only be the result of
    • 3. Food resource allocation patterns in non-reproductive males and females. I. 37genetic (σ2G) and environmental variation (σ2E), as is often suggested, but also ofvariation in stage of physiological development (σ2SPD): σ2P = σ2E + σ2SPD + σ2G (Luiting,1998). This may particularly be the case when comparing differences between lines orbreeds. Compared with genetically small animals, genetically large animals at a fixedbody weight grow faster, are leaner, and have better food efficiencies, not because oftrue superiority on these traits, but because the larger animals are physiologically lessmature than the smaller animals. According to Taylor (1980), the genetic variation can 2 2 2be further divided into scale effects (σ G-SIZE) and specific genetic factors (σ G-SGF): σ P= σ2E + σ2SPD + σ2G-SIZE + σ2G-SGF (Luiting, 1998). It is the specific genetic factors (SGFs)that cause the genotype to deviate from the standard expected values (Taylor, 1985). The strength of Taylor’s genetic size-scaling rules lies not only in its ‘unifieddescription of mammalian growth’, but also in its quantification of deviations from thestandard which are caused by the SGFs: it is these factors that are of genuine interestfor animal breeders. The deviations of the genotype from the standard expected valueshighlights its desirable or undesirable attributes (Taylor, 1985). Short-term selection responses result mainly from differences in genetic sizefactors and rarely from differences in SGFs. However, in the long term, SGFs becomemore important and statistically detectable, which means that long-term selection for asingle trait does more than merely change mature size: (1) accumulation of very smallsystematic differences in SGFs over generations may become finally detectable andstatistically significant, and (2) a further increase in genetic size may be limited by alimited uptake of resources from the environment such that SGFs become moreapparent, a situation described by the resource allocation theory (Beilharz et al., 1993).Further selection beyond this limit may change the metabolic resource situationconsiderably. This suggests that SGFs originate from variation in other traits than theones selected for (Taylor, 1982; Luiting et al., 1997). Consequently, scaling of growthand food intake curves according to Taylor (1980) is a tool to visualise (correlated)changes in the metabolic resource situation with selection.3. Material and methods3.1. AnimalsTwo lines of the Norwegian mouse selection experiment (e.g., Vangen, 1993) were used:a line selected for high litter size at birth (S-line) and a non-selected control line(C-line). In the 92nd generation (replicate 1), from 10 litters per line, 1 full-sib brother-sister pair was randomly chosen at weaning at 3 weeks of age and housed individually(i.e., 10 animals per sex per line). Replicate 1 started 1 day later for mice of the S-line,since S-line mice used in replicate 1 were born 1 day later than C-line mice. In the 95thgeneration (replicate 2), from 8 litters per line, 2 full-sib brothers-sisters pairs wererandomly chosen at weaning at 3 weeks of age and housed individually (i.e., 16 animalsper sex per line). The total data set comprised 104 animals with equal numbers per sexper line. Average total number of pups born in the 92nd and 95th generation was 21 and
    • 38 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knap22 in the S-line and 11 and 10 in the C-line, respectively. The selection response hasbeen at a plateaue for many generations (Vangen, 1993). The mice were housed in cages of 30 x 12.5 x 12.5 cm3 filled at the base withsawdust and had free access to pellet concentrate and water. The food contained12.6 kJ metabolisable energy per gram and 21% crude protein, as specified by theproducer. The mice originated from litters standardised at birth to 8 pups per litter,when larger than 8 pups. The light was left on 24 hours a day.3.2. Body weight and food intakeDuring 7 weeks (from 3 to 10 weeks of age), individual body weight (g) and foodconsumption (g) were measured from Monday to Friday daily in replicate 1, and 3 to 4times in replicate 2. From these data individual body weight gain (g) and cumulativefood intake (g) were calculated. Food consumption and body weight gain weresubsequently expressed on a daily basis. A Brody (1945) curve was fitted with the program SigmaPlot Scientific GraphingSystem (Jandel, 1992) to individual data on body weight against age; it is assumed thatmice at 3 weeks of age are in their ‘procreative phase’, i.e., beyond the point ofinflexion of the sigmoid-shaped growth curve:BWt = A (1 - e - k (t - t1*) ), (1a)where BW = body weight of the animal (kg) at age t (days from weaning), A = maturebody weight (kg), k = rate of maturation (per day), t = age (days from weaning), and t1*= a translation of equation (1a) along the X axis to complete the description of growth(days from weaning). A, k and t1* are parameters to be estimated. The same package was used to relate individual data of cumulative food intake toage according to a linear function by Parks (1982, p. 31), making the same assumption aswith equation (1a) and further assuming that the initial food intake at birth is negligiblysmall compared with mature food intake:CFIt = MFI (t - t2*), (2a)where CFIt = cumulative food intake of the animal (kg) at time t (days from weaning),MFI = maximum, mature daily food intake (kg/d), t = age (days from weaning), andt2* = X axis intercept (days from weaning). When body weight remains constant in anadult state, daily food intake is expected to be constant. MFI and t2* are parametersto be estimated.3.3. Genetic size scalingFollowing Taylor (1980), body weight (kg) and cumulative food intake (kg) of each animalwere scaled (i.e., divided) by its mature body weight (A (kg), equation 1), and age wasscaled by (t - 3.5)/A0.27, where (t - 3.5) is age in days from 3.5 days after conception;
    • 3. Food resource allocation patterns in non-reproductive males and females. I. 39gestation length is on average 19 days. Standardised body weight represents ‘degree ofmaturity’ (u) and standardised age represents ‘metabolic age’ (θ). To test for effects of SGFs, equations (1a) and (2a) were scaled according to Taylor(1980) in two steps by the individual estimates of mature body weight (A in kg). In thefirst step, all time parameters were scaled, which gives: 0.27 0.27BWt = A (1 - e - (k x A ) ((t – t1*) / A ) ), which can be written as 0.27BWt = A (1 - e - (k x A ) (θ - θ1*) ), and (1b)CFIt = MFI x A0.27 ((t - t2*) / A0.27), which can be written asCFIt = MFI x A0.27 (θ - θ2*). (2b)In the second step, in addition, BW and CFI were scaled: 0.27BWt / A = 1 - e - (k x A ) (θ - θ1*), which can be written as - (k x A0.27) (θ - θ1*)u=1-e , and (1c)CFIt / A = (MFI / A0.73) (θ - θ2*). (2c)3.4. Data analysisThe SAS program (Statistical Analysis System Institute, 1985) was used forstatistical analysis of all traits. The model used to describe all individual data was:Yijkl = µ + Ri + Lj + Gk + (RL)ij + (RG)ik + (LG)jk + (RLG)ijk + eijkl,where Yijkl = individual body weight, daily body weight gain, daily food intake, cumulativefood intake and all unscaled and scaled individual parameters estimated from equations(1) and (2), µ = overall mean, Ri = effect of replicate i (replicate 1, replicate 2),Lj = effect of line j (control, selected), Gk = effect of sex k (female, male),(RL)ij = interaction effect of replicate i by line j, (RG)jk = interaction effect of replicatei by sex k, (LG)jk = interaction effect of line j by sex k, (RLG)ijk = interaction effect ofreplicate i by line j by sex k, eijkl = error term of animal l, eijklNID(0, σ2e). Body weight,daily body weight gain, daily food intake and cumulative food intake were tested for 16data points which animals from replicate 1 and 2 had in common at equal ages (i.e., 21,22, 26, 28, 29, 33, 35, 36, 42, 43, 48, 50, 55, 57, 62, and 64 days of age).4. ResultsAverage curves for each sex in each line (adjusted for replicate effect) fittingequations (1a) and (2a) are given in Figure 1a. R2 values of individual curves fittingequation (1a) were 90% to nearly 100%; all R2 values of individual linear regressionsfitting equation (2a) were nearly 100%. Table 1 gives least squares means per sex per
    • 40 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knapline, adjusted for replicate effect, of the estimated parameters A, k, t1* (equation 1a),MFI and t2* (equation 2a). The Brody curve fitted the data very well, as indicated bythe R2 values. Therefore, the curve parameters could be estimated with reasonablyhigh accuracy’s. The coefficients of variation (i.e., the standard errors as a proportionof the associated estimate) among individual animals ranged from 1.70 to 2.81 for A,from 4.54 to 6.90 for k, and from 4.87 to 7.95 for t1*.Table 1 Least-squares means per sex per line (adjusted for effect of replicate) and standarderrors of the least-squares means, of mature body weight (A), maturation rate (k), translationparameter of equation (1a) (t1*), mature food intake (MFI), X axis intercept of equation (2a)(t2*), scaled k, scaled t1* (θ1*) , scaled MFI, and scaled t2* (θ2*). CF CM SF SM s.e. a b c dA (g) 31.2 38.4 43.0 51.5 0.877 a a a bk (per day) 0.0484 0.0526 0.0565 0.0736 0.00334t1* (days) 8.77a 11.2b 10.0ab 14.3c 0.697MFI (g) 4.86a 5.28b 6.63c 7.14d 0.0969 ab b c act2* (days) 21.8 21.7 22.3 22.0 0.0975Scaled k 0.0188a 0.0216ab 0.0239b 0.0328c 0.00136 ab ab a bθ1* 62.5 64.9 60.1 66.9 1.83Scaled MFI 0.0624a 0.0582b 0.0672c 0.0632a 0.00104 a b b cθ2 * 95.9 90.2 89.0 84.1 0.491 a,b,c,dC = control line, S = selection line, F = females, M = males; Values with different superscriptare significantly different, P<0.05. Data on daily body weight indicates that over the whole period, mice of the S-linewere heavier than mice of the C-line and males were heavier than females (P<0.001).From about 5 to 10 weeks of age, a significant interaction between line and sex (P<0.05)indicates that the difference between the sexes increased more in the S-line than inthe C-line. From about 5 to 6 weeks of age, the body weights were higher in replicate 1than in replicate 2 (P<0.05). During the growing period up to about 4 to 5 weeks of age,mice of the S-line grew faster per day than mice of the C-line and males grew fasterper day than females (P<0.01); afterwards the daily weight gain was very low and similarbetween both sexes and between both lines. Mature body weight (A) was significantly higher in S-line animals than in C-lineanimals and significantly higher in males than in females. Maturation rate (k) wassignificantly higher in males of the S-line than in S-line females and C-line animals. Thetime origin of equation (1a) (t1*) was significantly later in S-line animals than in C-lineanimals and significantly later in males than in females. Data on daily food intake indicates that S-line mice ate more than C-line mice(P<0.001); males ate more than females, but only before 55 days of age (P<0.001). Fromabout 3 to 6 weeks of age, animals ate more in replicate 1 than in replicate 2.Cumulative food intake was higher in S-line mice than in C-line mice, higher in males
    • 3. Food resource allocation patterns in non-reproductive males and females. I. 41 a 55 400 Cumulative food intake (g) 50 45 Body weight (g) 300 40 35 200 30 25 20 100 15 10 0 20 30 40 50 60 70 20 30 40 50 60 70 Age (d) Age (d) b Cumulative food intake (g) 55 400 50 45 Body weight (g) 300 40 35 200 30 25 20 100 15 10 0 80 130 180 230 80 130 180 230 Metabolic age (θ) Metabolic age (θ) c 1.1 10 Cumulative food intake / A Degree of maturity (u ) 8 0.9 6 0.7 4 0.5 2 0.3 0 80 130 180 230 80 130 180 230 Metabolic age (θ) Metabolic age (θ) CF CM SF SMFigure 1. Average curves fitting equations (1a) and (2a) in (a), (1b) and (2b) in (b), and (1c) and(2c) in (c) for each sex in each line. All curves are based on least-squares means of curveparameters adjusted for effect of replicate (N = 26): C = control line, S = selection line,F = female, M = male.
    • 42 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knapthan in females and higher in replicate 1 than in replicate 2 during the whole period(P<0.001). The estimate of mature daily food intake (MFI) was significantly higher in S-lineanimals than in C-line animals and significantly higher in males than in females. Asignificant interaction between line and replicate (P<0.05) indicates that the X axisintercept of equation (2a) (t2*) was significantly later in S-line animals than in C-lineanimals, but this was significant in replicate 1 only. The effect of scaling of the growth and food intake curves is shown in Figure 1b.Figure 1b shows average curves for each sex in each line (adjusted for replicateeffect) fitting equations (1b) and (2b), i.e., comparison of animals when accounting fordifferences in stages of physiological development. Figure 1c shows average curves foreach sex in each line (adjusted for replicate effect) fitting equations (1c) and (2c), i.e.,comparison of animals when accounting both for differences in physiologicaldevelopment and for scaling effects. The differences that remain after scaling bothaxes visualise the variation in SGFs. Least-squares means for each sex in each line (adjusted for replicate effect) of thescaled estimated parameters k, t1* (θ1*), MFI and t2* (θ2*) are given in Table 1. On a scaled basis, S-line animals matured faster (scaled k is higher) than C-lineanimals. A significant interaction between line and sex (P<0.05) indicates that malesmatured faster than females, but this is significant in the S-line only. The scaledtranslation parameter of equation (1a) (θ1*) was at a later metabolic age in S-line malesthan in S-line females. Mice of the S-line had a higher scaled MFI per metabolic day than mice of theC-line. A significant interaction between line and replicate (P<0.05) indicates that thisdifference was greater in replicate 2. Females had a higher scaled MFI per metabolicday than males. The scaled X axis intercept of equation (2a) (θ2*) was at a latermetabolic age in C-line animals than in S-line animals and later in females than in males.A significant interaction between line and replicate (P<0.01) indicates that the linedifference was greater in replicate 2.5. DiscussionIn accordance with other studies (e.g., Mgheni and Christensen, 1985), our results showthat selection for high litter size has increased daily and mature body weight in non-reproductive mice. During the growing period, selected mice grew faster than controlmice. Fitting of curves according to equation (1a) shows that selected males maturedfaster than selected females and C-line mice. In the present study, mature bodyweights were rather high compared with mature body weights of mouse lines reportedin the literature (e.g., Lang and Legates, 1969; Bakker, 1974). The parameter k in thepresent study falls in the range of 0.02 to 0.15 reported in the literature (e.g., Timonand Eisen, 1969; Eisen et al., 1969; Kastelic et al., 1996). Scaling of the time variables of equation (1a), resulting in equation (1b), shows theeffect of eliminating variation due to comparison of animals at different stages ofphysiological development. In other words, the variation between lines and sexes in 2 2 2Figure 1a results from aforementioned σ SPD + σ G-SIZE + σ G-SGF, while variation between
    • 3. Food resource allocation patterns in non-reproductive males and females. I. 43lines and sexes in Figure 1b results from σ2G-SIZE + σ2G-SGF (animals are compared in thesame environment; minor environmental effects can be assumed to be randomlydistributed among all individuals and are thus not a significant source of variationbetween lines and sexes). Since eliminating a source of variation is expected to makeanimals more similar, it is interesting to observe that the differences between linesand sexes at similar metabolical ages (Figure 2b) are larger than differences betweenlines and sexes when compared at similar ‘chronological’ ages (Figure 2a). An explanationmight be that the scaling parameter ‘mature body weight’ (A) adds some samplingvariance to the scaled X axis variable. Another explanation is that a negative covarianceexists between stages of physiological development and SGFs and/or scale effects. It has been debated whether the relationship between the time a species takes tomature in live weight and its mature weight raised to the power 0.27 similarly holdswithin a species and to sex differences within a breed or strain, or whether a differentrelationship of proportionality should be applied. Taylor (1968) found regressioncoefficients of time taken to mature on mature weight of 0.358 for strains and 0.343for sexes, and, although this was not significantly larger than 0.27, suggested that acoefficient of 1/3 might be a better value to adopt within species. Taylor and Fitzhugh(1971) conclude that ‘the genetic time scale of growth for species, breeds, sexes andindividuals can uniformly be taken as proportional to the 0.27 power of mature bodyweight’. Scaling growth and food intake variables in the present study with acoefficient of 1/3 (not presented) did not change the study’s conclusion. Scaling of growth variables of equation (1b), resulting in equation (1c), shows theeffect of additionally eliminating variation due to scale effects: the variation between 2lines and sexes in Figure 1c results from σ G-SGF only. Comparison of Figures 1a and 1bwith the fully scaled Figure 1c shows that differences between lines and sexes aregreatly decreased, although differences can still be recognised. Scaling of the timeorigin of the curve (t1*) has eliminated differences between lines but not betweensexes. However, scaling has increased differences between lines and sexes in rate ofmaturation (k): the difference between selected males and females remains, butselected mice mature faster than control mice when accounting for differences inmature size. This means that these animals reach a given proportion of their adult sizein a shorter fraction of their total lifespan. The results of the present study show that daily and mature food intake in non-reproductive mice are increased as correlated effects of selection for high litter size.Furthermore, male mice eat more per day than female mice during the first 8 weeks ofage. Cumulative food intake is higher in male mice than in female mice for the wholeperiod. It is very obvious from Table 1 that our estimates of t1* and t2* are not equal,although Parks (1982) implies that the time constant in his ‘ad libitum feeding function’(our t2*) equals Brody’s (1945) translation parameter (our t1*). There are three possiblereasons for this discrepancy. First, our assumptions of (i) the data points being beyondthe sigmoid-shaped growth curve’s point of inflexion and of (ii) initial food intake atbirth being negligibly small compared with mature food intake, may be invalid. Theseassumptions are necessary to justify the linear approach of equation (2a); theextremely high R2 values of the associated regressions suggest that both assumptions
    • 44 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knapseem valid in our data. Second, Parks’ (1982) implications may be wrong; the associatedreasoning in his book (page 31) is quite vague, using terms like ‘it seemed reasonable totry…’, ‘following the consequences (…) indicated the choice was reasonable’ and ‘afurther indication that t* (…) is Brody’s’. Third, our equations (1a) and (2a) were fittedseparately, without any constraint to arrive at the same t* estimate. This suggests acombined analysis of the body weight and food intake data as is presented in theaccompanying paper (Chapter 4). Scaling of time variables of equation (2a), resulting in equation (2b), shows that, asfor the aforementioned effect on growth curves, variation in cumulative food intakebetween lines and sexes increases after scaling, which can be explained in a similar wayas for body weight. Scaling of food intake variables of equation (2b), resulting inequation (2c) reduces the variation between lines and sexes greatly. However,differences between lines can still be recognised here as well. Estimates of maturefood intake remain higher in S-line mice compared with C-line mice after adjustmentfor differences in mature size. However, before scaling, males have higher mature dailyfood intake than females, while after scaling females have higher estimates of maturefood intake per metabolic day than males. According to Taylor (1980), the differences that remain after scaling the growthand food intake curves are a reflection of SGFs. The results indicate that long-termselection for high litter size in mice has done more than only change the mature size. Ithas influenced the metabolic resource situation. Other long-term selection studiesshowing significant changes in the size-scaled shape of the growth curve andconsequently the presence of SGFs, are reported by McCarthy and Siegel (1983; miceand broiler chickens), Thompson et al. (1985, sheep), Rickleffs (1985, quails and broilerchickens) and Siegel and Dunnington (1987, broiler chickens) with selection on growthrate. However, no references could be found where growth patterns have beeninvestigated in a population selected for litter size. Taylor chose mature body weight, A, to represent genetic body size, since it is morestable than any other measure (Taylor, 1985). According to Taylor (1985), ‘a definitionof mature body weight that meets most purposes is the body weight of a normallygrown, skeletally mature, normally active adult animal maintained in a state of bodyweight equilibrium on a standard diet, in a thermo-neutral, disease-free environmentwith, or adjusted to, a chemical body fat of (…) 15%’. Others use the point at whichprotein accretion ceases as an estimate of mature body weight (e.g., Owens et al.,1995). Considering its rather detailed definition, A is not an ideal scaling parameter forpractical use; an obvious disadvantage is that it can only be measured when theasymptotic weight (adjusted for fat percentage) is attained. Mice following the normalgrowth pattern reach the phase of skeletal maturation from the 7th to the 20th week ofage, and the phase of full maturation from the 26th to the 52nd week of age (Malik,1984). Since growth curves in the present study are fitted to ages of only 10 weeks,the mature body weights, A, are likely underestimated. Bünger and Schönfelder (1984)showed in laboratory mice that the asymptotic values of sigmoid growth curves, asestimated from body weight measurements that run up to first mating, mayunderestimate the eventually realised body weight by 30%. Further experiments shouldaim at collecting data up to later ages.
    • 3. Food resource allocation patterns in non-reproductive males and females. I. 45 According to the resource allocation theory, in a limited environment, an optimalsituation exists where all resources available are diverted to each resource demandingprocess such that the overall homeostatic balance (i.e., fitness) is optimised (Beilharzet al., 1993). Selection for a single trait, however, may lead to the situation wheredisproportionally many resources are diverted towards the trait selected for, leavingthe animal lacking in ability to respond adequately to other demands. Since theselection response in litter size in the Norwegian mouse selection experiment has beenat a plateaue for many generations (Vangen, 1993), it may be expected that theselected mice are limited by their environment. In the present study, SGFs aredetected for both body weight and food intake, which suggests that selection forincreased litter size has disproportionally changed the resource allocation pattern.Since body weight can be described as a function of cumulative food intake, as is welldescribed by Parks (1982), SGFs for body weight may be related to SGFs for foodintake. The accompanying paper (Chapter 4) aims at investigating whether thedifferences in body weight can be explained by differences in food intake.AcknowledgementsThis study has been supported by a grant from the Norwegian Research Council,project number 114258/111, ‘Consequences of selection for high litter size for theability to cope in a stressful environment’. Kari Kjus is gratefully acknowledged forcarrying out the Norwegian mouse selection experiment and her help in providing andmaintaining the mice of the present study.
    • 46 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. KnapReferences• Bakker, H., 1974. Effect of selection for relative growth rate and body weight of mice on rate, composition and efficiency of growth. PhD Thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Beilharz, R.G., Luxford, B.G., Wilkinson, J.L., 1993. Quantitative genetics and evolution: is our understanding of genetics sufficient to explain evolution? J. Anim. Breed. Gen. 110:161-170.• Brody, S., 1945. Bioenergetics and growth. Reinhold, New York.• Bünger, L. and Schönfelder, E., 1984. Zur lebensleistung wachstumsselektierter labormausweibchen – wachstumsverlauf. Probleme der angewandten Statistik 11:185-196.• De la Fuente, L.F. and San Primitivo, F., 1985. Selection for large and small litter size of the 1st 3 litters in mice. Gen. Sel. Evol. 17:251-264.• Eisen, E.J., Lang, B.J., Legates, J.E., 1969. Comparison of growth functions within and between lines of mice selected for large and small body weight. Theor. Appl. Gen. 39:251- 260.• Jandel, 1992. SigmaPlot Scientific Graph System users manual, version 5.0. Jandel Scientific Inc., San Rafael CA, USA.• Kastelic, M., Šalehar, A., Drobnic, M., Kovac, M., 1996. Mature size and maturing rate in RoC57BL/6 and RoNMRI lines of mice. J. Anim. Breed. Genet. 113:545-551.• Lang, B.J. and Legates, J.E., 1969. Rate, composition and efficiency of growth in mice selected for large and small body weight. Theor. Appl. Genet. 39:306-314.• Luiting, P., 1998. The role of genetic variation in feed intake and its physiological aspects: results from selection experiments. In: Regulation of Feed Intake. Proceedings of the 5th Zodiac Symposium, Wageningen, The Netherlands (ed. D. van der Heide, E.A. Huisman, E. Kanis, J.W.M. Osse and M.W.A. Verstegen), pp. 75-87. CAB International, Wallingford.• Luiting, P., Vangen, O., Rauw, W.M., Knap, P.W., Beilharz, R.G., 1997. Physiological consequences of selection for growth. 48th Annual Meeting of the EAAP, Vienna, Austria.• Malik, R.C., 1984. Genetic and physiological aspects of growth, body composition and feed efficiency in mice: a review. J. Anim. Sci. 58:577-590.• McCarthy, J.C. and Siegel, P.B., 1983. A review of genetical and physiological effects of selection in meat-type poultry. Anim. Breed. Abstr. 51:87-94.• Mgheni, M. and Christensen, K., 1985. Selection experiment on growth and litter size in rabbits. III. Two-way selection response for litter size. Acta Agric. Scand. 35:287-294.• Narayan, A.D. and Rawat, S., 1986. Effect of selection for litter size on body weight in mice. Indian Vet. Med. J. 10:82-87.• Ogink, N.W.M., 1993. Genetic size and growth in goats. PhD thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Owens, F.N., Gill, D.R., Secrist, D.S., Coleman, S.W., 1995. Review of some aspects of growth and development of feedlot cattle. J. Anim. Sci. 73:3152-3172.• Parks, J.R., 1982. A theory of feeding and growth in animals. Springer-Verlag, New York.• Rickleffs, R.E., 1985. Modification of growth and development of muscles of poultry. Poultry Sci. 64:1563-1576.• Statistical Analysis Systems Institute, 1985. SAS user’s guide: statistics. Version 5. Statistical Analysis Systems Institute Inc., Cary, NC.• Siegel, P.B. and Dunnington, E.A., 1987. Selection for growth in chickens. CRC Crit. Rev. Poultry Biol. 1:1-24.
    • 3. Food resource allocation patterns in non-reproductive males and females. I. 47• Taylor, St C.S., 1980. Live-weight growth from embryo to adult in domesticated mammals. Anim. Prod. 31:223-235.• Taylor, St C.S., 1982. Theory of growth and feed efficiency in relation to maturity in body weight. Proc. 2nd World Congr. Gen. Appl. Livest. Prod., Madrid, Spain, Vol. 5, p 218-230.• Taylor, St C.S., 1985. Use of genetic size-scaling in evaluation of animal growth. J. Anim. Sci. 61: (Suppl. 2) 118-141.• Taylor, St C.S. and Fitzhugh, H.A., 1971. Genetic relationships between mature weight and time taken to mature within a breed. J. Anim. Sci. 33:726-731.• Taylor, St C.S. and Murray, J.I., 1987. Genetic aspects of mammalian growth and survival in relation to body size. Academic Press, London.• Thompson, J.M., Parks, J.F., Perry, D., 1985. Food intake, growth and body composition in Australian Merino sheep selected for high and low weaning weight. 1: Food intake, food efficiency and growth. Anim. Prod. 40:55-70.• Timon, V.M. and Eisen, E.J., 1969. Comparison of growth curves of mice selected and unselected for postweaning gain. Theor. Appl. Gen. 39:345-351.• Vangen, O., 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.
    • 4 FOOD RESOURCE ALLOCATION PATTERNS IN NON-REPRODUCTIVE MALES AND FEMALES. II. TRENDS IN BODY WEIGHT AGAINST FOOD INTAKE W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. KnapAbstract In the accompanying paper, specific genetic factors for body weight andfood intake were identified in non-reproductive male and female mice of a line selectedfor high litter size at birth (average of 22 born per litter) and a non-selected controlline (average of 10 born per litter). The existence of these factors is indicated byvariation in efficiency parameters such as growth efficiency and maintenancerequirements. Residual food intake (RFI) and Parks’ estimates of growth efficiency(AB) and maintenance requirements (MEm) were used to quantify these factors. In thegrowing period, females had a higher RFI (are less efficient) than males. At maturity,selected mice had higher RFI than control mice and selected females had higher RFIthan selected males. AB was higher in selected mice than in control mice, and higher inmales than in females. MEm was higher in selected mice than in control mice, and higherin females than in males. The results indicate the existence of specific genetic factorsfor both growth efficiency and maintenance requirements. Selected females mayincrease RFI in adult state to anticipate the metabolically stressful periods ofpregnancy and lactation, to support a genetically highly increased litter size.Keywords: food intake, growth, litter size, selectionBased on Animal Science 71, W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knap,Differences in food resource allocation in a long-term selection experiment for litter size inmice. II. Developmental trends in body weight against food intake, 39-47, ©2000 British Societyof Animal Science, with permission from British Society of Animal Science.
    • 50 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knap1. IntroductionIn the accompanying paper (Chapter 3) it was shown that growth and food intake curvesagainst time, of non-pregnant individuals resulting from a long-term selectionexperiment for litter size in mice, differ significantly when standardised by maturebody weight. Hence this indicates the presence of line specific genetic factors (SGFs).These results suggest that selection for high litter size has disproportionally changedthe resource allocation pattern. The presence of SGFs for both body weight and foodintake suggest an interrelationship that may be described by efficiency parameterssuch as growth efficiency and maintenance requirements. Calculation of residual food intake (RFI) and the method of Parks (1982) toestimate maintenance requirements and growth efficiency are tools to quantify theseSGFs. RFI is defined as the part of the food intake that is unexplained by maintenanceand production, or in other words, as the difference between the food that isconsumed by an animal and its consumption as predicted from a model involving itsgrowth and maintenance requirements. Variation in RFI can be caused by variation inpartial efficiencies for maintenance and growth and by variation in metabolic (i.e., fooddemanding) processes not included in the model, such as physical activity, responses topathogens and responses to stress (Luiting and Urff, 1991). Parks’ (1982) method involves estimation of growth efficiency and maintenancerequirements per metabolic kg by fitting curves of body weight against cumulative foodintake. The aim of the present study is to investigate the metabolic resource situation innon-reproductive males and females in a long-term selection experiment for litter sizein mice. The present paper aims to quantify and explain the line differences in SGFsthat have been detected in the accompanying paper (Chapter 3).2. Material and methodsThe experimental set-up, the conditions under which the mice were kept and themeasurement methods for body weight and food intake are described in theaccompanying paper (Chapter 3). Briefly, two mouse lines of the Norwegian mouseselection experiment (Vangen, 1993) were used: a line selected for high litter size atbirth (S-line) and a non-selected control line (C-line). The experimental populationconsisted of 10 and 16 animals per sex per line originating from two generations: the92nd (replicate 1) and the 95th (replicate 2), respectively. Body weight and food intakewere measured from 3 to 10 weeks of age, for 5 days/week in replicate 1 and 3 to 4days/week in replicate 2.2.1. Residual food intakeThe equation used to estimate RFI for each animal was based on the following multiplelinear regression of food intake on metabolic body weight and body weight gain in the
    • 4. Food resource allocation patterns in non-reproductive males and females. II. 51control line (within replicate):FCi(C) = b0(C) + (b1(C) × BWi(C)0.75) + (b2(C) × BWGi(C)) + ei(C), (1)where FCi(C) = food consumption of mouse i of the C-line (g/day), BWi(C)0.75 = metabolicbody weight of mouse i of the C-line (kg0.75), BWGi(C) = body weight gain of mouse i ofthe C-line (g/day), b0(C) = C-line population intercept, b1(C), b2(C) = C-line population partialregression coefficients representing maintenance requirements per metabolic kg andrequirements for growth, respectively, and ei(C) = the error term, representing RFI ofmouse i of the C-line (g/day). Equation (1) was fitted per day from 3 to 10 weeks of age. Subsequently, within replicate, RFI of S-line individuals was estimated as: ˆ ˆ 0.75 ˆRFIi(S) = FCi(S) - { b 0(C) + ( b 1(C) × BWi(S) ) + ( b 2(C) × BWGi(S))}, (2)where RFIi(S) = residual food intake of mouse i of the S-line (g/day), FCi(S) = foodconsumption of mouse i of the S-line (g/day), BWi(S)0.75 = metabolic body weight ofmouse i of the S-line (kg0.75), and BWGi(S) = body weight gain of mouse i of the S-line(g/day); b0(C), b1(C) and b2(C) are the parameters described in (1). This was done using theestimates from each day of measurements from 3 to 10 weeks of age. The experimental period was subsequently divided, based on daily growth ratespresented in the companion paper (Chapter 3), into a ‘growing period’, i.e., from 3 to 6weeks of age, and an ‘adult period’, i.e., from 6 to 10 weeks of age. Equation (1) wasfitted for the growing period and the adult period from cumulated data on growth anfood intake per animal over these periods. Metabolic body weight of the growing periodwas estimated as the average of metabolic body weights at 3, 4, 5 and 6 weeks of age;metabolic body weight of the adult period was estimated as the average of metabolicbody weights at 6, 7, 8, 9 and 10 weeks of age. Finally, equation (1) was fitted for thetotal period with metabolic body weight estimated as the average of the week valuesfrom 3 to 10 weeks of age. RFI for each animal in the C-line equalled its error term ei(c) in (1) and RFI of S-lineindividuals was estimated as in equation (2). This implies that the average RFI of allC-line mice equalled 0. RFI was consequently estimated per day, for the growing period,for the adult period and for the total period from 3 to 10 weeks of age.2.2. Parks’ curvesFollowing Archer and Pitchford (1996), modified Parks’ (1982) curves were fitted withthe program SigmaPlot Scientific Graphing System (Jandel, 1992) to individual data onbody weight against cumulative food intake from 3 to 10 weeks of age:BWt = A (1 - e - B(CFIt + FI0) ), (3)where BWt = body weight of the individual (g) at age t (days from weaning),CFIt = cumulative food intake (g) at age t (d from weaning; at weaning CFI = 0),
    • 52 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. KnapA = mature (adult, asymptotic) body weight (g), B = rate of maturation of body weightwith respect to food intake (per g), FI0 = a translation of equation (3) along the X axisto complete the description of growth (g). A, B and FI0 are parameters to be estimated.The modification of the Parks’ curve involves the inclusion of FI0 to avoid fixing thecurve through any point (Archer and Pitchford, 1996). The same package was used to relate individual data on cumulative food intake toage according to a linear function by Parks (1982, p. 31). To ensure that the food intakerepresented food intake at maturity, data were restricted to 60 days of age and older(Archer and Pitchford, 1996):CFIt = MFI (t - t*), (4)where CFIt = cumulative food intake of the animal (g) at time t (days from weaning, > 60days of age), MFI = maximum, mature daily food intake (g/day), t = age (days fromweaning, > 60 days of age); and t* = X axis intercept (days from weaning). MFI and t*are parameters to be estimated. Following Taylor (1980), body weight (kg) and cumulative food intake (kg) of eachanimal were scaled (i.e., divided) by its mature body weight (A in kg, equation 3), andage was scaled by (t - 3.5)/A0.27, where (t - 3.5) is age in days from 3.5 days afterconception (for the justification of why the value of 3.5 was chosen, see Chapter 3);gestation length is on average 19 days. Standardised body weight represents ‘degree ofmaturity’ (u) and standardised age represents ‘metabolic age’ (θ). Scaling of equation(3) gives: - (B x A) {(CFIt/A) + (FI0/A)}BWt/A = 1 - e , which can be written as - (B x A) {(CFIt/A) + (FI0/A)}u=1-e (5)The two parameters B and FI0 are now scaled to B x A and FI0/A; Parks (1982) definedA x B as the growth efficiency parameter (AB). Equation (4) was scaled in the same way as equation (2a) in the accompanying paper(Chapter 3):CFIt/A = (MFI/A0.73) {(t/A0.27) - (t*/A0.27)}, which can be written asCFIt/A = (MFI/A0.73) (θ - θ*) (6)The two parameters MFI and t* are now scaled to MFI/A0.73 and θ*; when MFI isexpressed in kJ metabolic energy by multiplying it with the metabolisable energycontent of the food (12.6 kJ ME per g), MFI/A0.73 represents the adult maintenancerequirements per metabolic kg (MEm).2.3. Data analysesThe SAS program (Statistical Analysis Systems Institute, 1985) was used forstatistical analyses of RFI and all estimated parameters. The model used to describe
    • 4. Food resource allocation patterns in non-reproductive males and females. II. 53the individual data was:Yijkl = µ + Ri + Lj + Gk + (RL)ij + (RG)ik +(LG)jk +(RLG)ijk + eijkl,where Yijkl = individual daily RFI, RFI in the growing period, the adult period and thetotal period and all unscaled and scaled parameters estimated from equations (3) and(4), µ = overall mean, Ri = effect of replicate i (replicate 1, replicate 2), Lj = effect ofline j (control, selected), Gk = effect of sex k (female, male), (RL)ij = interaction effectof replicate i by line j, (RG)jk = interaction effect of replicate i by sex k,(LG)jk = interaction effect of line j by sex k, (RLG)ijk = interaction effect of replicate iby line j by sex k, eijkl = error term of animal l, eijklNID(0,σ2e). Phenotypic correlationsbetween RFI in the growing period, the adult period and the total period, and A, MFI,AB and MEm were estimated adjusted for replicate, line and sex effects and theirinteractions.3. Results3.1. Residual food intakeFigure 1 shows the average daily RFI for each sex in each line from 3 to 10 weeks ofage. Results are presented for each replicate. Trends are described by linear lines forC-line mice and logarithmic lines for S-line mice. R2 values of the multiple regressionsaccording to equation (1) per day were in the range of 42% to 86% in replicate 1 and47% to 83% in replicate 2. Least-squares means of RFI in the growing period, the adultperiod and the total period for each sex in each line (adjusted for effect of replicate)are presented in Table 1. R2 values of the multiple regressions according to equation (1)per period in replicate 1 and replicate 2 were 88% and 81% for the growing period, 74%and 66% for the adult period, and 77% and 75% for the total period, respectively. Figure 1 shows that in C-line animals a trend in RFI from 3 to 10 weeks of age wasabsent: average daily RFI in C-line animals hovered around 0. The reason for this isthat equation (1) was based on all C-line animals. Hence the average RFI of the C-linepopulation was 0 and therefore the average values of males and females of the C-linewere symmetrical around 0. Figure 1 shows an increasing trend in S-line animals: ingeneral RFI of S-line mice was lower than RFI of C-line mice during the first daysafter weaning and higher during the last weeks. This shift occurred somewhat sooner in replicate 2 (around 4.5 weeks of age) thanin replicate 1 (around 5.5 weeks of age). Therefore, in the growing period from 3 to 6weeks of age, S-line mice had a lower RFI than C-line mice in replicate 1, although thisis not significant, and a significantly higher RFI in replicate 2. In the adult period andover the total period, S-line mice had significantly higher RFI than C-line mice, i.e.,they were less food efficient. Figure 1 shows furthermore that RFI was consistently higher in females than inmales. In the growing period, a significant interaction between line and sex (P<0.05)indicates that this sex difference was significantly greater in the S-line than in the
    • 54 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. KnapC-line. In the adult period, a significant interaction between line and sex (P<0.01)indicates that this sex difference was significant in S-line mice only. A significantinteraction between sex and replicate (P<0.05) indicates that the difference in RFIbetween males and females was significantly greater in replicate 1 in the growing periodand in the total period. 2.0 2.0 CF 1.5 1.5 CM Residual food intake (g) 1.0 1.0 SF 0.5 0.5 SM 0.0 0.0 -0.5 -0.5 -1.0 -1.0 -1.5 -1.5 21 26 31 36 41 46 51 56 61 66 71 21 26 31 36 41 46 51 56 61 66 71 Age (days)Figure 1 Average daily residual food intake from 3 to 10 weeks of age for each sex in eachreplicate of each line. The period before the vertical line (drawn at 6 weeks of age) is thegrowing period; after the line is the adult period (N = 10 in replicate 1, N = 16 in replicate 2).C = control line; S = selection line; F = female; M = male. Linear trend lines are fitted to data onC-line mice; logarithmic trend lines are fitted to data on S-line mice.3.2. Parks’ curvesAverage curves for each sex in each line (adjusted for replicate effect) fittingequations (3) and (5) are given in Figures 2a and 2b, respectively. R2 values of individualcurves were 90% to nearly 100%. All R2 values of individual linear regressions fittingequation (4) were 100%. Least squares means of A, B, FI0 (equation 3), MFI, t* (equation 4), scaled B (AB),scaled FI0 (equation 5), scaled MFI multiplied by the energy content of the food (MEm)and scaled t* (equation 6) for each sex in each line (adjusted for replicate effect), arepresented in Table 1. Parks’ growth curves fitted the data very well, as indicated by theR2 values. Therefore, the curve parameters could be estimated with reasonably highaccuracies. The coefficients of variation (i.e., the standard errors as a proportion ofthe associated estimate) among individual animals ranged from 1.48 to 2.45 for A, from4.73 to 5.55 for B, and from 4.49 to 6.51 for FI0. S-line individuals had a significantly higher mature body weight (A) than C-linemice and A was significantly higher in males than in females. These estimates of A
    • 4. Food resource allocation patterns in non-reproductive males and females. II. 55were, as expected, very similar to the estimates of A according to equation (1a) in theaccompanying paper (Chapter 3). S-line males had a significantly higher maturation rateof body weight with respect to food intake (B) than S-line females. Estimates of Bwere very similar for C-line males and females and lay in between the estimates ofS-line males and females. A significant interaction between line and sex (P<0.05)indicates that the time origin of equation (3) (FI0) was later in females than in malesbut this difference was significant in the S-line only.Table 1 Least-squares means for each sex in each line (adjusted for effect of replicate) andstandard errors of the least-squares means, of daily residual food intake (RFI) in the growingperiod, the adult period and the total period, Parks’ parameter estimates of mature body weight(A), rate of maturation of body weight with respect to food intake (B), translation parameter ofequation (3) (FI0), mature food intake (MFI), X axis intercept of equation (4) (t*), scaled B(growth efficiency, AB), scaled FI0, scaled MFI multiplied by the metabolisable energy contentof the food (maintenance requirements per metabolic kg, MEm), and scaled t* (θ*). CF CM SF SM Standard error ª b a bRFI growing period (g) 0.0859 -0.0859 0.212 -0.180 0.0553 ª a b cRFI adult period (g) 0.0822 -0.0822 0.811 0.271 0.0664 ª b c aRFI total period (g) 0.0827 -0.0827 0.550 0.195 0.0544A (g) 30.5ª 37.8b 41.9c 50.5d 0.748 ab ab a bB (per g) 0.0107 0.0104 0.0098 0.0115 0.000544 ª ab c bFI0 (g) 57.2 51.2 67.2 46.4 3.02MFI (per d) 4.93ª 5.22b 6.78c 6.87c 0.0924 ª b c bt* (d) 22.5 21.3 23.9 21.1 0.438AB 0.321ª 0.381b 0.408b 0.577c 0.0174 ª ab c bscaled FI0 1.86 1.35 1.60 0.926 0.0683MEm (kJ/kg) 857ª 774b 923b 814c 13.0θ* 100ª 91.6b 95.2c 84.2d 1.10C = control line; S = selection line; F = females; M = males. Values with different superscripts aresignificantly different (P<0.05). Estimates of mature food intake (MFI) were significantly higher in S-line micethan in C-line mice. MFI was significantly higher in males than in females, though thiswas significant in the C-line only. Furthermore, MFI was significantly higher in replicate2 than in replicate 1 (P<0.001). The estimates according to equation (4) were verysimilar to the estimates according to equation (2a) in the accompanying paper (Chapter3), although in the present experiment, the linear function related cumulative foodintake to age greater than 60 days. Differences between males and females in MFIaccording to equation (4) were smaller than in MFI according to equation (2a) in theaccompanying paper (Chapter 3); differences between S-line males and females were nolonger significant. A significant interaction between line and replicate (P<0.05) indicatesthat the X axis intercept of equation (4) (t*) was significantly later in S-line animals
    • 56 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knapthan in C-line animals in replicate 1 but no significant line difference existed inreplicate 2. The t* was significantly later in females than in males. Estimates of scaled rate of a 55 maturation with respect to food intake 50 (growth efficiency, AB) were 45 Body weight (g) 40 significantly higher in mice of the S-line 35 30 than in mice of the C-line. Males had 25 higher AB than females; a significant 20 15 interaction between line and sex 10 (P<0.01) indicates that this difference 0 50 100 150 200 250 300 350 was significantly greater in the S-line Cumulative food intake (g) than in the C-line. Scaled FI0 was b significantly later in C-line animals than 1.1 Degree of maturity (u ) in S-line animals and significantly later 0.9 in females than in males. 0.7 S-line mice had significantly higher scaled MFI multiplied by the 0.5 metabolisable energy content of the 0.3 food (maintenance requirements per 0 2 4 6 8 metabolic kg, MEm) than C-line mice; a Cumulative food intake / A significant interaction between line and replicate (P<0.01) indicates that this CF CM SF SM was mainly caused by replicate 1. MEmFigure 2 Average curves fitting equations 3 was significantly higher in females than(a) and 5 (b) for each sex in each line. All in males. Scaled t* (θ*) was later in C-curves are based on least-squares means of line mice than in S-line mice, though acurve parameters for each sex in each line, significant interaction between line andadjusted for effect of replicate (N = 26). C replicate (P<0.01) indicates that this= control line; S = selection line; F = female; was significant in replicate 2 only. θ*M = male. was later in females than in males.Table 2 Phenotypic correlations between residual food intake (RFI) in the growing period, theadult period and the total period, and Parks’ parameter estimates of mature body weight (A),mature food intake (MFI), growth efficiency (AB) and maintenance requirements per metabolickg (MEm), adjusted for replicate, line and sex. RFI growing period RFI adult period RFI total periodA -0.01 0.17 0.05MFI 0.41 *** 0.82 *** 0.78 ***AB 0.22 * -0.11 0.21 *MEm 0.41 *** 0.43 *** 0.72 ****P<0.05; ***P<0.001. Table 2 presents phenotypic correlations (adjusted for replicate, line and sex)between RFI in the growing period, the adult period and the total period, with A, MFI,AB and MEm. The correlations between RFI in the growing period and RFI in the adult
    • 4. Food resource allocation patterns in non-reproductive males and females. II. 57period (r = 0.47), between RFI in the growing period and RFI in the total period(r = 0.77) and between RFI in the adult period and RFI in the total period (r = 0.87)were positive and highly significant (P<0.001). RFI in none of the periods wassignificantly correlated with A. RFI in the growing period and in the total period butnot in the adult period, was significantly positively correlated with AB. RFI in allperiods was significantly positively correlated with both MFI and MEm. The correlationbetween RFI and MFI was highest in the adult period and lowest in the growing period.The correlation between RFI and MEm was highest in the total period.4. DiscussionIn the accompanying paper, significant differences in SGFs for both body weight andfood intake were identified in non-reproductive male and female mice between a lineselected for high litter size at birth and a non-selected control line (Chapter 3). Theseresults lead to the question: is the genetic change in the body weight curve againsttime caused by the genetic change in the food intake curve against time; or in otherwords, has selection for high litter size disproportionally changed the resourceallocation pattern? Furthermore, scaling of time variables with A0.27 (Chapter 3) is notvery well verified within species. Taylor (1985) suggested, therefore, to use cumulativefood intake as a time-scale for growth instead of chronological age, and the subsequentscaling of cumulative food intake with A. For these reasons, in the present paper, RFIand Parks (1982) curves of body weight against cumulative food intake were studied inthe same lines. Both tools describe the relationship between body weight and food intake with asingle parameter during growth or at maturity. RFI must be interpreted as a conversionparameter (i.e., output/input) and is estimated during growth and at maturity. Parksestimate of growth efficiency (AB) is an efficiency parameter (i.e., input/output) and isestimated during growth. Parks estimate of maintenance requirements per metabolickg (MEm) is a conversion parameter and is estimated at maturity. The higher maturation rate independent of mature body weight in the selectionline, as presented in the accompanying paper (Chapter 3), suggests a higher growthefficiency in the selection line. Indeed Parks estimate of AB was significantly higher inthe selection line, especially in males. Parks estimate of AB represents the amount of food required for the first unit ofbody weight gain when there is no body weight to maintain and all food is directedtowards growth. As such it should be most closely related to the RFI at the first dayof measurement (which is not the first day of growth but as close as we can get), albeitan efficiency parameter versus a conversion parameter. Indeed, RFI was significantlylower in the selection line at the first day (and more strongly so than on subsequentdays), which corresponds with the higher AB in the selection line. Furthermore, AB represents the constant rate of decrease in growth efficiencywith increasing degree of maturity (Parks, 1982). This is caused by an increase in foodrequirements for maintenance relative to the food requirements for growth, and by anincrease in the food requirements per unit growth caused by an increase in lipid-to-protein ratio, when the animal matures. Because AB was higher in the selection line, its
    • 58 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knapmaintenance requirements per metabolic kg must have been higher and/or the lipid-to-protein ratio in its body gain must have increased faster than in the control line. Theformer corresponds to the higher MEm estimated in the selection line (especially infemales), although this is an estimate at maturity which could be different from thatat an immature stage. The phenotypic correlation between AB and MEm (adjusted forline, sex and replicate) points in the same direction (r = 0.39, P < 0.01). Differences ingrowth composition between these lines will be investigated in forthcomingexperiments. Figure 3 shows the estimated decrease in growth efficiency with increasing degreeof maturity (AB × (1-u), where u = body weight/A (Chapter 3)) for the two lines andsexes (adjusted for replicate effect). It shows a similar pattern to that found for RFIduring growth (Figure 1): S-line males are somewhat more efficient during growth andshow a somewhat lower RFI (although not significant) than the other three line-by-sexcombinations. Also within line, sex and replicate, the pattern of variation in AB × (1- u)and RFI are similar: phenotypic correlations between AB × (1- u) and RFI, adjusted forline, sex and replicate range, over time, between 0 and -0.28 (becoming significant at-0.20 for P<0.05). The fact that RFI during growth in 0.40 the females of the selection line is positive is caused by the early change in 0.30 CF daily RFI from negative to positive. A AB * (1 - u ) CM negative RFI in S-line animals is caused 0.20 SF by an overestimation of food 0.10 SM requirements by the model, i.e., as 0.00 expected from food requirements of the control population on which the -0.10 20 40 60 80 model is formed. Correspondingly, a Age (days) positive RFI in S-line animals is caused by an underestimation of foodFigure 3 Decrease in growth efficiency with requirements by the model.increasing degree of maturity (AB x (1- u)) Overestimation of food requirementsfor each sex in each line, adjusted for for S-line animals during the first daysreplicate effect (N = 26). u = body after weaning is caused byweight/A; C = control line; S = selection line; overestimation of food required forF = female; M = male. growth: S-line animals grow more efficient than C-line animals. Thesubsequent underestimation of food requirements for S-line animals occurs when foodrequirements for maintenance are underestimated: MEm is higher in S-line animals thanin C-line animals. The change in RFI in S-line animals from negative to positive occurswhen the higher MEm dominates the higher AB. Because MEm in S-line females is muchhigher than in S-line males, and AB in S-line females is lower than in S-line males, thischange occurs much earlier in the females than in the males, leading to an overall higherRFI in the growing period. Presumably, this also explains the positive sign of the phenotypic correlations ofRFI in the growing period with AB and MEm. Archer and Pitchford (1996) found a
    • 4. Food resource allocation patterns in non-reproductive males and females. II. 59negative phenotypic correlation between AB and RFI in the first two weeks afterweaning and no significant correlation thereafter. They also found no significantcorrelation between MEm and RFI in the first two weeks after weaning and a positivecorrelation thereafter. This could indicate a difference in the point where thedominance of AB changes into the dominance of MEm (namely a bit later). The latter isquite possible, considering the large environmental effect on MEm. In the adult period, S-line mice had a higher RFI than C-line mice and S-linefemales had a higher RFI than S-line males. Since growth is virtually absent atmaturity, the differences in RFI are largely caused by differences in maintenancerequirements per metabolic kg. This corresponds very well with the higher MEmestimate in the selection line, especially in the females. Furthermore, RFI in the adultperiod also has a strong positive phenotypic correlation with MEm within line, sex andreplicate (and, obviously, there is no significant correlation between RFI in the adultperiod and AB). Since tissues with high protein or high lipid levels have differentmaintenance requirements, line differences in body composition may explain part of theline difference in adult RFI. Differences in adult body composition between the lineswill be investigated in forthcoming experiments. However, Archer and Pitchford (1996)found that variation in body composition in adult mice accounted for only a very smallpart (less than 9%) of variation in RFI. The connection between Taylors (1980) genetic size scaling rules and RFI in adultanimals is described by Luiting et al. (1997): adjustment for adult size in equation (1) isaccomplished by the inclusion of metabolic body weight (BW0.75) as a covariate in themodel, which is similar to the inclusion of A0.73 in the case of adult animals.Consequently, both the estimates of adult RFI (i.e., mature food intake in g/dayadjusted linearly for mature metabolic weight) and Parks mature food intake scaledaccording to Taylor (1980) as presented in equation (6) (i.e., mature food intake ing/day divided by mature metabolic weight), are estimates of mature food intakeindependent of mature size. With RFI, adjustment for mature size is accomplished bylinear regression rather than by division, which is statistically a better approach (Packardand Bordman, 1988). In growing animals, however, the linear adjustment with the covariate BW0.75 is notthe same as adjustment for adult size A0.73, and comparisons are made at the samechronological age. Hence, some adjustments for mature size and physiological age wouldbe needed. The first could be taken care of by including A0.73 as a covariate into themodel, and the latter by measuring food intake and weight over the entire growthperiod up to the stage when mature size is reached or by including some measurementof stage of development in the model (given the terms already present in the model, itis not clear what this measurement should be). However, the correlations between RFIand mature size were very close to zero and non-significant in our data (Table 2) andthe same holds for the correlations between RFI and stage of development (calculatedas W/A) in Archer and Pitchford (1996). This suggests that the phenotypicadjustments for size and stage of development will not be very crucial but they may bewhen the variation between animals in these traits is larger. Our estimate of growth efficiency AB in the S-line males is particularly highcompared with the ones presented by Parks (1982) and Archer and Pitchford (1996) in
    • 60 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. Knapmice (see Figure 4). However, AB of the S-line males is predicted very well by thelinear regression of AB on A based on these literature data and our own data (withoutthe S-line males) as given in Figure 4: they have a high AB because of their very high Ain comparison to the other populations. The linear relation between AB and A showsfurthermore that variation in this scaled efficiency parameter can still be explained byvariation in mature size: AB is not independent of A. Replicate 1 was conducted from June to August 1997 while replicate 2 wasconducted from March to May 1998. There has not been any climate regulation ineither of the replicates, but temperature and humidity have been registered inreplicate 2: temperature was on average 20.3°C (s.e. 0.63°C) and average humidity was61.6% (s.e. 4.8%). It is likely that the temperature was higher and humidity lowerduring the summer months. Indeed, daily mature food intake and MEm were lower inreplicate 1. This could be the reason for the later appearance of the point where thedominance of AB changes into the dominance of MEm as can be seen from the change insign of RFI (hence a lower RFI in the growing period in replicate 1 than in replicate 2). This seems probable because of the large effect of 60 MEm on food efficiency both during 55 growth and at maturity. Because many 50 maintenance processes are very 45 dependent on environmental factors and AB 40 are sometimes only expressed under 35 certain circumstances, line by replicate 30 interactions could be expected. 25 From the point of view of resource 25 35 45 55 allocation, animals with low RFI may be A particularly short in resources left toFigure 4 Estimates of growth efficiency respond adequately to unexpected(AB) in relation to estimates of mature body stresses and diseases. Laying hensweight (A): triangles: Archer and Pitchford selected for high RFI were better(1996); circles: Parks (1982); squares: adapted to stressors such as relocation,present study; open: Females; solid: Males. A high temperatures and ACTH than henslinear regression line is fitted through all selected for low RFI (Luiting et al.,data excluding selection-line males of the 1994; Bordas and Minvielle, 1997). It ispresent study (AB = 6.47 + 0.92A, interesting that particularly adult 2R = 0.60). females of the selection line have a very high RFI, since it is these animals thatcan express the trait their genotype has been selected for: a high litter size at birth.This higher RFI may therefore anticipate the metabolically stressful periods ofpregnancy and especially lactation. These periods may be expected to considerablychange the physiological resource demand, since the S-line females have to support agenetically highly increased litter size. The question is whether this can be supportedby an increase in food intake during these periods or whether the RFI will dropconsiderably. Forthcoming research will aim to further investigate this.
    • 4. Food resource allocation patterns in non-reproductive males and females. II. 61AcknowledgementsThis study has been supported by a grant from the Norwegian Research Council,project number 114258/111, ‘Consequences of selection for high litter size for theability to cope in a stressful environment’. Kari Kjus is gratefully acknowledged forcarrying out the Norwegian mouse selection experiment and her help in providing andmaintaining the mice of the present study.
    • 62 W.M. Rauw, P. Luiting, M.W.A. Verstegen, O. Vangen and P.W. KnapReferences• Archer, J.A., Pitchford, W.S., 1996. Phenotypic variation in residual food intake of mice at different ages and its relationship with efficiency of growth, maintenance and body composition. Anim. Sci. 63:149-157.• Bordas, A. and Minvielle, F., 1997. Réponse à la chaleur de poules pondeuses issues de lignées sélectionnées pour une faible (R-) our forte (R+) consommation alimentaire résiduelle. Gen. Sel. Ev. 29:279-290.• Jandel, 1992. SigmaPlot Scientific Graph System users manual, version 5.0. Jandel Scientific Inc., San Rafael CA, USA.• Luiting, P., Decuypere, E., De Groot, P.N., Buyse, J., Broom, G., 1994. Selection for feed efficiency and consequences for stress susceptibility. 45th Annual Meeting of the EAAP, Edinburgh, Scotland UK.• Luiting, P. and Urff, E.M., 1991. Optimization of a model to estimate residual feed consumption in the laying hen. Livest. Prod. Sci. 27:321-338.• Luiting, P., Vangen, O., Rauw, W.M., Knap, P.W., Beilharz, R.G., 1997. Physiological consequences of selection for growth. 48th Annual Meeting of the EAAP, Vienna, Austria.• Packard, G.C. and Bordman, T.J., 1988. The misuse of ratios, indices, and percentages in ecophysiological research. Phys. Zool. 61:1-9.• Parks, J.R., 1982. A theory of feeding and growth of animals. Springer-Verlag, Berlin.• Statistical Analysis Systems Institute, 1985. SAS user’s guide: statistics. Version 5. Statistical Analysis Systems Institute Inc., Cary, NC.• Taylor, St. C.S., 1980. Genetic size-scaling rules in animal growth. Anim. Prod. 30:161-165.• Taylor, St. C.S., 1985. Use of genetic size-scaling in evaluation of animal growth. J. Anim. Sci. 61: (Suppl. 2) 118-141.• Vangen, O., 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.
    • 5 BODY COMPOSITION IN NON- REPRODUCTIVE ADULT MALES AND FEMALES W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. LuitingAbstract Earlier studies showed that adult mice from a line selected for high littersize (S-line), in particular females, had higher residual food intake (RFI) than micefrom a non-selected control line (C-line). It was suggested that this increase in RFI, inparticular the mature selected females, may anticipate the metabolically stressfulperiods of pregnancy and lactation. The present study investigated whether also bodyprotein and lipid stores at maturity have been affected as a correlated effect ofselection, for later support of fetus growth and lactation. Furthermore, part of theobserved differences between individuals in RFI may be attributable to differingproportions of body protein and lipid. For these reasons, differences in bodycomposition at maturity between males and females of the S-line and the C-line wereinvestigated. Relative lipid mass was similar for C-line animals and S-line females; S-linemales had a significantly lower relative lipid mass. Males had a higher relative proteinmass than females, in particular S-line males. The results show that body composition inadult non-reproductive females has not been affected as a correlated effect ofselection for high litter size. Furthermore, the results suggest that the high leancontent in S-line males may explain part of the high RFI compared with C-line animals.Body composition in S-line females does not likely explain the high RFI compared withS-line males and C-line animals. Factors other than protein and lipid levels must beresponsible for the differences found between the lines and sexes in RFI.Keywords: mice, selection, litter size, residual food intake, body compositionBased on W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting, Body composition in non-reproductive adult males and females in a long-term selection experiment for litter size in mice,submitted.
    • 64 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting1. IntroductionChapter 4 showed that mature non-reproductive individuals (6 to 10 wk of age), and inparticular females, from a line selected for more than 90 generations for high littersize at birth (S-line) have a significantly higher residual food intake (RFI) than mice ofa non-selected control line (C-line). RFI of males and females of the C-line was –0.0822and 0.0822 g/d and of males and females of the S-line 0.271 and 0.811 g/d,respectively. The method to calculate RFI in the C- and S-line population has been extensivelydescribed in Chapter 4. Briefly, RFI in non-reproductive animals was calculated,according to Luiting and Urff (1991), by regressing food intake on metabolic bodyweight and body weight gain (Chapter 4). The partial regression coefficient for foodconsumption on metabolic body weight is defined as the average maintenancerequirement per unit of metabolic body weight and the partial regression coefficient offood consumption on body weight gain is defined as the average food requirement forgrowth. Residual food intake was defined as the difference between the food thatactually is consumed by an animal and its predicted consumption from observed bodyweight and growth of the C-line population. Therefore, the average RFI of the C-linepopulation equaled zero. Variation in RFI can be caused by variation in requirements for maintenance andgrowth, and by variation in food demanding processes not explicitly considered in themodel, such as activity and response to stress. Since growth is virtually absent atmaturity, variation in maintenance requirements is the major cause of variation in RFIin adult animals (Luiting, 1990). Indeed, adult mice with higher RFI had significantlyhigher estimates of maintenance requirements (according to Parks (1982); r = 0.43,adjusted for line and sex), while no significant relationship was found between RFI andestimates of growth efficiency (according to Parks (1982); r = -0.11, adjusted for lineand sex; Chapter 4). In Chapter 4 it was suggested that the increase in RFI at maturity in particularlythe S-line females may anticipate the periods of pregnancy and lactation which willrequire a high metabolic rate. These periods will considerably increase the physiologicalresource demand. This will especially be the case in S-line females since they have tosupport a litter that has practically been doubled in size by selection: the averagenumber of live-born pups is about 20 in the S-line and 10 in the C-line. Therefore,before and during pregnancy, S-line females may increase their protein and lipid storesfor later support of fetal growth and lactation. The objective of this study was todetermine whether body composition at maturity has been changed as a consequence ofselection for high litter size. Furthermore, since tissues with high protein or high lipid levels have differentmaintenance requirements, line and sex differences in body composition may explainpart of the variation in RFI (Webster, 1985; Luiting, 1990). Therefore, the secondobjective of this study was to investigate if the observed differences between theC- and S-lines in body composition of non-reproducing adult males and females areconsistent with expectations from previously documented differences between thelines in RFI.
    • 5. Body composition in non-reproductive adult males and females 652. Materials and methodsTwo lines of a Norwegian mouse selection experiment (e.g., Vangen, 1993, 1999) wereused: a line selected for high litter size at birth (S-line) and a non-selected control line(C-line). In the 101st generation, from each of 30 litters per line, one full-sib brother-sister pair was randomly chosen at weaning. All animals were housed in pairs with miceof the same sex and the same line. Average litter size at birth in the 101st generationwas 10 in the C-line and 21 in the S-line. The mice were housed in 30 x 12.5 x 12.5 cm3 cages bedded with sawdust. Animalshad free access to pellet concentrate and water. The food contained 12.6 kJ ME pergram and 21% crude protein, as specified by the supplier. The mice originated fromlitters standardized at birth to eight pups per litter, when larger than eight pups. Thelight was left on 24 hours a day.2.1. Body compositionBody composition was measured individually at 65 days of age. Animals were deniedaccess to food 10 hours before killing but had free access to water; animals were killedwith CO2. The mice were weighed just before and just after this procedure. Deadanimals were stored at -20°C. Samples were prepared by boiling carcasses individuallyin a glass jar in water for 10 minutes; subsequently, carcasses were minced individuallyin a kitchen blender (Moulinex Illico). Minced samples were stored at -20°C. Totalnitrogen (N) was estimated by the Kjeldahl method; total protein content wascalculated as N x 6.25. Total lipid was estimated by ether extraction pretreated withHCl.2.2. Data analysesAbsolute protein mass was calculated as protein content per gram sample multiplied bythe body weight of the dead mouse; absolute lipid mass was calculated in a similar way.Percent protein was calculated as protein content per gram sample multiplied by thebody weight of the live mouse and expressed as a percentage of live body weight; lipidpercentage was calculated in a similar way. The SAS program (SAS, 1985) was used for statistical analyses of all traits. Themodel used to describe the data was:Yijk = µ + Li + Gj + (LG)ij + eijk,where µ = overall mean, Li = effect of line i (control, selected), Gj = effect of sex j(female, male), (LG)ij = interaction effect of line i by sex j, eijk = error term of animal k,eijkNID(0, σ2e). Yijk denotes all traits tested with this model, all as measured on animal kof line i and sex j: live body weight, absolute lipid and protein mass, and lipid andprotein percentages.
    • 66 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting 25 r P-value 20 Lipid percentage CF 0.50 0.0051 CM 0.62 0.0003 15 SF -0.04 0.8179 SM 0.30 0.1085 10 5 20 25 30 35 40 45 50 55 23 22 r P-value Protein percentage 21 -0.65 0.0001 CF 20 -0.88 0.0001 CM 19 SF -0.42 0.0223 18 SM -0.47 0.0083 17 16 15 20 25 30 35 40 45 50 55 Live body w eight (g)Figure 1 Phenotypic relation between lipid percentage and protein percentage, and live bodyweight in non-reproductive males and females at 65 days of age. C = control line; S = selectionline; F = female; M = male. 23 22 Protein percentage 21 r P-value 20 CF -0.83 0.0001 19 CM -0.76 0.0001 18 SF -0.64 0.0001 17 SM -0.60 0.0001 16 15 5 10 15 20 25 Lipid percentageFigure 2 Phenotypic relation between protein percentage and lipid percentage in non-reproductive males and females at 65 days of age. C = control line; S = selection line; F = female;M = male.
    • 5. Body composition in non-reproductive adult males and females 673. ResultsAverage live body weight, absolute lipid mass, absolute protein mass, lipid percentage,and protein percentage of non-reproductive males and females at 65 days of age arepresented in Table 1. S-line mice were significantly heavier than C-line mice and maleswere significantly heavier than females. Absolute lipid mass was significantly lower in C-line females compared with C-linemales and S-line individuals. Lipid percentage was significantly lower in S-line malescompared with S-line females and C-line individuals. Absolute protein mass wassignificantly higher in S-line mice than in C-line mice and significantly higher in malesthan in females. Protein percentage was significantly higher in males than in females,and significantly higher in S-line males than in C-line males (Table 1).Table 1 Means and standard errors of live body weight, absolute lipid mass, absolute proteinmass, lipid percentage and protein percentage of non-reproductive males and females at 65 daysof age. CFa CMa SFa SMaLive body weight (g) 27.49 ± 0.38w 34.02 ± 0.66x 38.80 ± 0.65y 47.23 ± 0.73z y z zAbsolute lipid mass (g) 3.35 ± 0.23 4.56 ± 0.28 4.75 ± 0.25 4.56 ± 0.22zAbsolute protein mass (g) 5.02 ± 0.05w 6.42 ± 0.08x 7.05 ± 0.11y 9.55 ± 0.13zLipid percentage (%) 12.06 ± 0.71y 13.20 ± 0.62y 12.26 ± 0.63y 9.60 ± 0.40zProtein percentage (%) 18.31 ± 0.17x 18.95 ± 0.17y 18.20 ± 0.15x 20.25 ± 0.13zWithin a row, means without a common superscript letter differ (P < 0.05). aC = control line;S = selection line; F = female; M = male; N=30 for each line in each sex.Figure 1 presents the phenotypic relation between live body weight, and lipid andprotein percentage and Figure 2 presents the phenotypic relation between lipidpercentage and protein percentage, at 65 days of age for each sex in each line. InC-line animals and S-line males, heavier animals had a higher lipid percentage, thoughthis was significant for C-line mice only. For S-line females a slightly negative, non-significant correlation was found between live body weight and lipid percentage (Figure1). Heavier animals had a lower protein percentage; this correlation was stronger inC-line animals (Figure 1). Animals with a higher lipid percentage had a lower proteinpercentage (Figure 2).4. DiscussionLipid and protein percentages in the present study are very similar to the average lipidand protein percentage reported by Bakker (1974) in a randombred control line (11.89%and 18.61% at 8 weeks of age, respectively). He observed that from 3 to 15 weeks ofage females generally had a higher lipid and lower protein percentage than males; thiswas non-significant at 8 weeks of age. In the study of Lang and Legates (1969), femalesgenerally had a significantly higher relative lipid mass and lower relative protein massthan males from 3 to 8 weeks of age. Furthermore, Hayes and Eisen (1979) observed at63 days of age a significantly higher lipid (+7.38%) and significantly lower protein
    • 68 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luitingcontent (-3.77%) in females compared with males. The results of the present study arein line with these results from the literature: females had a lower protein percentagethan males and S-line females had a higher lipid percentage than S-line males. The negative phenotypic correlation between percentage lipid and protein in thepresent study supports observations by Lang and Legates (1969) and Bakker (1974),who found a phenotypic correlation, within line, sex and age, of –0.36 (age from 3 to 15weeks) and –0.73 (age from 3 to 8 weeks), respectively. Chapter 3 showed that selection for high litter size has resulted in a correlatedincrease in mature body size: estimates of mature body weight are about 35% larger inS-line mice than in C-line mice. When animals with different mature sizes are comparedat the same chronological age, differences may be found that result from variation indegree of maturity. Compared with genetically small animals, genetically large animals ata fixed age have a higher relative protein mass and a lower relative lipid mass becausethe larger animals are physiologically less mature than the smaller animals (Taylor,1980). However, Chapter 3 also showed that estimates of rate of maturation weresignificantly higher in S-line mice than in C-line mice, significantly so in S-line males.Degree of maturity in body weight (i.e., body weight divided by the estimate of maturebody weight (A)) at 65 days of age, as estimated from data of Chapter 3, is 90.9% at207 metabolic days (i.e., age/A0.27) in C-line females, 91.1% at 195 metabolic days inC-line males, 94.0% at 190 metabolic days in S-line females, and 96.5% at 180 metabolicdays in S-line males. At 65 days of age, S-line mice have a higher degree of maturity inbody weight than C-line animals, significantly so in S-line males (P < 0.001). Therefore,in the present study, degree of maturity likely does not explain line and sexdifferences in body composition. It can be suggested that the fetus acts as a parasite that would deplete the damfrom her energy stores (Pine et al., 1994). Indeed, several studies in rats indicate thatfood restricted dams transfer a similar percent energy to their offspring as ad libitumfed animals, which will deplete the dams energy stores more than in ad libitum fedanimals (e.g., Pine et al., 1994; Luz and Griggio, 1996). Young and Halliday (1997)observed that the amount of maternal body protein that was transferred across theplacental membrane to the fetuses was not altered by reducing the maternal foodintake by two-thirds. This means that restricted fed animals have less food energyavailable for maintenance and their own development in absolute quantity comparedwith ad libitum fed animals. Usually, during the first half of pregnancy, body proteinand lipid content is increased for later support of fetal growth and lactation (Naismithet al., 1982; Luz and Griggio, 1996). Since S-line females have to support a litter sizethat has been highly increased by artificial selection, the present study investigatedwhether body composition at maturity has been affected as a correlated effect ofselection for high litter size, to sufficiently support the offspring during pregnancyand lactation. Results of the present study indicate that this is not the case: there areno differences in body stores of protein and lipid between C-line and S-line adult non-reproductive females. Forthcoming research will investigate food resource allocationpatterns and body composition in lactating C- and S-line females. Food eaten by an animal is used for several functions, such as maintenance, growthand (re)production. A major fraction of the difference in energy requirements between
    • 5. Body composition in non-reproductive adult males and females 69non-reproductive individuals, when kept under uniform nutritional and environmentalconditions, is caused by differences in metabolic body weight and body weight gain. Yet,some variation in energy requirements remains unexplained. This fraction of foodrequirements independent of body weight and production is called residual food intake(RFI) (Luiting, 1990). In Chapter 4 it was shown that long-term selection for litter sizein mice has resulted in animals with higher RFI in adult state, especially in females. In adult, non-reproducing animals, variation in RFI can be mainly explained byvariation in maintenance requirements (Luiting, 1990). Since energy requirements formaintenance per metabolic kilogram tend to be lower in fat animals than in lean animals(Webster, 1985), part of the observed differences between individuals in RFI may beattributable simply to differing proportions of protein and lipid in the body (Luiting,1990). Maintenance of body fat requires little energy because it is relativelymetabolically inactive, whereas body protein is continually degraded into amino acidsand resynthesized. Estimated energy costs of fat and protein turnover are 2% to 3%(Katz and Rognstad, 1976) and 15% to 25% (MacRae and Lobley, 1986) of basal energyexpenditure, respectively. Consequently, energy requirements for maintenance mayshow protein turn-over-related variation among animals as a result of variation in bodycomposition (Knap, 1996). For this reason, in the present study, it was investigated if the observeddifferences between the C- and S-lines in body composition of non-reproducing adultmales and females are consistent with the expectations from previously documenteddifferences between the lines in RFI. It is then expected that S-line animals, especiallyfemales of this line, are leaner than C-line animals. In the study of Katle (1991), abdominal fat content made a significant but smallcontribution to the variation in RFI: at 65 weeks of age, laying hens selected for threegenerations for high RFI (as a fraction of E(FI)) were leaner and produced more heatfor maintenance per unit of metabolic body size than hens from a low RFI line. Theresults were supported by Katle and Nordli (1992). Also, two lines of laying hens thathad been divergently selected for RFI for four generations by Luiting et al. (1991)differed significantly in fat content: the low RFI line contained 3.4% more fat.Furthermore, El-Kazzi et al. (1995) found at 52 weeks of age a significantly lowerfatness in laying hens selected for 17 generations for high RFI compared with a lowRFI line; males from the high RFI line had extreme low fatness. These resultssupported observations at earlier generations (e.g., Zein-el-Dein et al., 1985; Bordasand Mérat, 1991). In the study of Archer and Pitchford (1996) estimates of body water content, whichis positively related to the amount of lean tissue, at 29 weeks of age showed significantpositive correlations with RFI in mice older than 7 weeks. These correlations weresimilar to those between body water content and estimates of maintenancerequirements. However, the effect of body composition on RFI appeared to be small.Mérat et al. (1980) observed a phenotypic correlation of -0.16 between RFI andabdominal adipose tissue in laying hens. The correlation between RFI and body composition does not necessarily hold forgrowing animals, since fat is energetically more expensive to deposit than lean(Webster, 1985). In laying hens, Bentsen (1983) observed a positive phenotypic
    • 70 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luitingcorrelation between RFI and abdominal fat from 16 to 40 weeks of age, which becamenegative from 40 to 66 weeks of age. Archer et al. (1998) reported a phenotypiccorrelation between RFI and proportion of body fat in mice of –0.02 during a post-weaning period (approximately 4 to 6 weeks of age) and of –0.23 during a mature period(approximately 21 to 23 weeks of age). In growing bulls of dual purpose breeds, Jensenet al. (1992) observed that phenotypically fat animals tended to be more inefficient;the phenotypic correlation between RFI and fat percentage was 0.22. In that study,RFI estimated with and without correction for carcass composition were closelycorrelated, indicating that consideration of carcass composition in estimating RFI ingrowing bulls adds very little information. Mrode and Kennedy (1993) reported that RFIin immature Yorkshire, Landrace and Duroc boars was positively phenotypicallycorrelated with backfat thickness adjusted to 90 kg live weight (r = 0.20). Similarresults were found by Johnson et al. (1999) in Large White boars: the phenotypiccorrelation between RFI and backfat at approximately 176 days of age was 0.33. So itdepends on composition of gain during development and even more on body compositionwhat will dominate the deviation in RFI. In conclusion, it is generally observed that mature animals with high RFI are leanerthan animals with low RFI, but that the effect of body composition on RFI appears tobe small. Body fat content is often strongly associated with body weight and bodyweight gain (Luiting, 1990). Archer et al. (1998) reported a phenotypic correlation of0.58 between proportion of body fat and body weight in mature mice. Thus, the modelwhen calculating RFI accounts for an important part of the variation in bodycomposition. Interestingly, in the present study, this is true for C-line animals only. The results of the present study suggest that the higher lean content in S-linemales may explain part of the higher RFI found in these animals as compared withC-line animals. Body composition in S-line females, however, is not likely to explain theparticularly high RFI as compared with S-line males and C-line animals. This means thatfactors other than protein and lipid levels must be responsible for the differencesfound between the lines and sexes in RFI. Several studies have indicated that a higherRFI is related to higher activity levels (e.g., Luiting et al., 1991; De Haer et al., 1993).Differences in activity levels between adult non-reproductive C- and S-line females areinvestigated in Chapter 6: indeed, in response to a novel environment, S-line femalesshow more activity than C-line females.AcknowledgementsThis study has been supported by a grant from the Norwegian Research Council,project number 114258/111 and by the S. and M. Berges Forskningsfond. Kari Kjus isgratefully acknowledged for carrying out the Norwegian mouse selection experimentand her help in providing and maintaining the mice of the present study. Prof Knut Hoveand Anna Haug are gratefully acknowledged for their suggestions on the method forobtaining the body composition samples. The Laboratory for Analytical Chemistry (LAK)is acknowledged for the analysis of the samples. This manuscript has partly beenwritten at the Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria(I.N.I.A.) in Madrid, Spain.
    • 5. Body composition in non-reproductive adult males and females 71References• Archer, J.A. and Pitchford, W.S., 1996. Phenotypic variation in residual food intake of mice at different ages and its relationship with efficiency of growth, maintenance and body composition. Anim. Sci. 63:149-157.• Archer, J.A., W.S. Pitchford, T.E. Hughes, and P.F. Parnell. 1998. Genetic and phenotypicrelationships between food intake, growth, efficiency and body composition of mice post weaning and at maturity. Anim. Sci. 67:171-182.• Bakker, H. 1974. Effect of selection for relative growth rate and body weight of mice on rate,composition and efficiency of growth. PhD thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Bentsen, H.B. 1983. Genetic variation in feed efficiency of laying hens at constant body weight and egg production. II: Sources of variation in feed consumption. Acta Agric. Scand. 33:305- 320.• Bordas, A., and P. Mérat. 1991. Sélection divergente pour la consommation alimentaire "résiduelle" de la poule en période de ponte: réponse au taux protéique de l’aliment. Genet. Sel. Evol. 23:249-256.• De Haer, L.C.M., P. Luiting, and H.L.M. Aarts. 1993. Relations among individual (residual) feed intake, growth performance and feed intake pattern of growing pigs in group housing. Livest. Prod. Sci. 36:233-253.• El-Kazzi, M., A. Bordas, G. Gandemer, and F. Minvielle. 1995. Divergent selection for residual food intake in Rhode Island Red egg-laying hens: gross carcass composition, carcass adiposity and lipid contents of tissues. Br. Poult. Sci. 36:719-728.• Hayes, J.F., and E.J. Eisen. 1979. Environmental maternal influences on body composition in mice selected for body weight. Theor. Appl. Gen. 55:209-223.• Jensen, J., Mao, I.L., Andersen, B.B., 1991. Genetic parameters of growth, feed intake, feed conversion and carcass composition of dual-purpose bulls in performance testing. J. Anim. Sci. 69:931.• Johnson, Z.B., J.J. Chewning, and R.A. Nugent. 1999. Genetic parameters for production traits and measures of residual feed intake in Large White Swine. J. Anim. Sci. 77:1679-1685.• Katle, J., 1991. Selection for efficiency of food utilisation in laying hens: causal factors for variation in residual food consumption. Br. Poult. Sci. 32:955-969.• Katle, J., and H. Nordli. 1992. Variation in residual food consumption between cocks in a highly productive egg-laying strain. Acta Agric. Scand. 42:20-26.• Katz, J., and R. Rognstad. 1976. Futile cycles in the metabolism of glucose. In: B. Horecker, and E. Stadman, (eds) Current Topics in Cellular Regulation No 10. Academic Press, New York, p. 271-274.• Knap, P.W. 1996. Stochastic simulation of growth in pigs: protein turn-over-dependent relations between body composition and maintenance requirements. Anim. Sci. 63:549-561.• Lang, B.J., and J.E. Legates. 1969. Rate, composition and efficiency of growth in mice selected for large and small body weight. Theor. Appl. Gen. 39:306-314.• Luiting, P. 1990. Genetic variation of energy partitioning in laying hens: causes of variation in residual feed consumption. World’s Poult. Sci. J. 46:133-152.• Luiting, P., J.W. Schrama, W. Van der Hel, E.M. Urff, P.G.J.J. Boekholt, E.M.W Van den Elsen, and M.W.A. Verstegen. 1991. Metabolic differences between White Leghorns selected for high and low residual feed consumption. 12th Symp. Energy Metab. Farm Anim., Kartause Ittingen, Switzerland.
    • 72 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting• Luiting, P. and Urff, E.M., 1991. Optimization of a model to estimate residual feed consumption in the laying hen. Livest. Prod. Sci. 27:321-338.• Luz, J. and Griggio, M.A., 1996. Distribution of energy between food-restricted dams and offspring. Ann. Nutr. Metab. 40:165-174.• MacRae, J.C., and G.E. Lobley. 1986. Interactions between energy and protein. In: L.P. Milligan, W.L. Grovum, and A. Dobson (eds) Control of Digestion and Metabolism in Ruminants. Prentice- Hall, Englewood Cliffs, New Jersey.• Mérat, P., A. Bordas, and F.H. Ricard. 1980. Composition anatomique, production d’oeufs et efficacité alimentaire de poules pondeuses. Corrélations phénotypiques. Ann. Génét. Sél. Anim., 12:191-200.• Mrode, R.A., and B.W. Kennedy. 1993. Genetic variation in measures of food efficiency in pigs and their genetic relationships with growth rate and backfat. Anim. Prod. 56:225-232.• Naismith, D.J., Richardson, D.P., Pritchard, A.E., 1982. The utilization of protein and energy during lactation in the rat, with particular regard to the use of fat accumulated in pregnancy, Br. J. Nutr. 48:433-441.• Parks, J.R., 1982. A theory of feeding and growth of animals. Springer-Verlag, Berlin.• Pine, A.P., Jessop, N.S., Oldham, J.D., 1994. Maternal protein reserves and their influence on lactational performance in rats. Br. J. Nutr. 71:13-27.• SAS, 1985. SAS user’s guide: statistics. Version 5. Sas Institute Inc., Cary, NC, USA.• Taylor, St C.S. 1980. Live-weight growth from embryo to adult in domesticated mammals. Anim. Prod. 31:223-235.• Vangen, O. 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.• Vangen, O. 1999. Long term selection for litter size in mice (>100 generations); correlated responses and biological constraints. 50th Annual Meeting of the EAAP.• Webster, A.J.F., 1985. Differences in the energetic efficiency of animal growth. J. Anim. Sci. 61: (Suppl. 2) 92-103.• Young, M. and Halliday, D., 1997. A direct method for investigating the transfer of endogenous nitrogen from maternal body protein to fetal protein. Placenta 18:469-472.• Zein-el-Dein, A., A. Bordas, and P. Mérat. 1985. Sélection divergente pour la composante "résiduelle" de la consommation alimentaire des poules pondeuses: effets sur la composition corporelle. Arch. Geflügelk. 49:158-160.
    • 6 BEHAVIOURAL STRATEGIES IN NON-REPRODUCTIVE ADULT FEMALES W.M. Rauw, P. Luiting, M. Bakken, T. Schuurman, C.J.M. de Veer and O.VangenAbstract In a previous study it has been shown that adult, non-reproductive femalemice from a line selected for high litter size at birth (S-line) have higher residual foodintake (RFI) than females of a non-selected control line (C-line). Several studies haveindicated that a higher RFI is related to a higher activity level. Differences in activitymay suggest underlying differences in coping strategies. To investigate whether copingstrategies have been affected as a correlated effect of selection for high litter size,48 non-reproductive mature C- and S-line females were twice subjected to an openfield, seven times to a maze, once to a social confrontation and twice to a runway test.In the second open-field test, S-line females crossed more squares than C-line femalesand were less reserved to enter the centre of the open field. In the maze test, S-linefemales showed more locomotion activity than C-line females. They encountered dead-ends more often and returned more often to the starting area in the first maze test.No differences were found in their reaction towards an extra-maze cue change. In thesocial confrontation test, S-line females showed more locomotion activity, were lessimmobile and investigated the floor and opponent less than C-line females. S-linefemales ran faster in both runway tests than C-line females. These results suggestthat the novelty response of S-line females is more dominated by an active coping stylethan that of C-line females. However, more tests, such as additional behavioural tests,physiological tests and neurobiological tests are required to be more conclusive onwhether selection for high litter size has resulted in mice that adopt the active copingstrategy.Keywords: selection, litter size, coping behaviour, resource allocationBased on Applied Animal Behaviour Science 66, W.M. Rauw, P. Luiting, M. Bakken, T. Schuurman,C.J.M. de Veer and O. Vangen, Behavioural differences in non-reproductive adult females in along-term selection experiment for litter size in mice, 249-262, ©2000 Elsevier Science B.V.,with permission from Elsevier Science.
    • 74 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. Vangen1. IntroductionIn Chapter 4 it was shown that mature non-reproductive female mice of a line selectedfor high litter size have higher residual food intake (RFI) than females of a non-selected control line. RFI is defined as the part of the food intake that is unexplainedby food requirements for maintenance and production, or in other words, as thedifference between the food that is consumed by an animal and its consumption aspredicted from a model involving its growth and maintenance requirements. Variation inRFI can be caused by variation in partial efficiencies for maintenance and growth andby variation in metabolic food demanding processes not included in the model, such asbehavioural activities, responses to pathogens and responses to stress. Since growth isvirtually absent at maturity, the differences in RFI are mainly explained by differencesin maintenance requirements (Luiting, 1990). In a divergent selection experiment on RFI in laying hens, Luiting et al. (1991)observed higher activity-related heat production in the high RFI (low efficiency) linethan in the low RFI (efficient) line: more than 50% of the variation in RFI could beexplained by variation in physical activity. Findings by Morrison and Leeson (1978) andBraastad and Katle (1989) support these results: inefficient laying hens were moreactive than efficient hens. Variation in the amount of food intake activity accountedfor 44% of the variation in RFI in pigs: pigs with a low RFI visited the feed hopper lessoften and spent less time eating per day than pigs with a high RFI (De Haer et al.,1993); these results are supported by Von Felde et al. (1996). Differences in activity levels suggest underlying differences in coping strategies.Coping can be defined as a behavioural and physiological reaction to aversive situationsand aims at retaining or re-establishing homeostasis (Wechsler, 1995; Koolhaas et al.,1997). Behavioural strategies have been extensively described in mice selectedbidirectionally for attack latency, particularly in males (e.g., Benus, 1988; Sluyter et al.,1996) and more recently also in females (Compaan et al., 1993; Benus and Röndigs,1996). The correlated responses in a variety of other behavioural traits indicate theexistence of a general underlying control mechanism determining complex behaviouralstrategies (Sluyter et al., 1995). Consequently, reciprocal selection for relatedbehavioural traits, such as nest-building behaviour (Sluyter et al., 1995), active shockavoidance acquisition (Driscoll et al., 1990) and susceptibility to apomorphine (Cools etal., 1990) have resulted in similar reciprocal correlated responses in coping behaviour.Large individual differences in the behavioural and physiological reactions to stressorsare not restricted to rodent species. Evidence for the existence of different copingstyles has been obtained in pigs (Hessing et al., 1993), dairy cows (Hopster, 1998) andgreat tits (Verbeek et al., 1996). In general, two coping strategies are distinguished: active and passive copers, whichadopt different strategies towards a wide variety of environmental challenges(Koolhaas et al., 1997). Active copers tend to actively manipulate events, whereaspassive copers tend to switch to passivity. In social confrontations, active copersrespond by attacking the opponent or fleeing from a physically stronger opponent(fight/flight reaction), whereas passive copers respond with ‘freezing behaviour’ orimmobility (conservation/withdrawal strategy). In non-social situations, the behaviour
    • 6. Behavioural strategies in non-reproductive adult females 75of active copers is less affected by changes in the environment. Active copers tend todevelop routine behaviour; their behaviour is more intrinsically organised. Thebehaviour of passive copers is more dependent on external stimuli and hence theirbehaviour is more flexible (Benus, 1988). Active and passive copers differ furthermorefor several physiological and neurobiological characteristics (Benus, 1988). The aim of the present study was to investigate whether selection for high littersize at birth in mice has resulted, as a correlated effect, in a higher occurrence ofbehavioural characteristics indicative of an active coping style. Therefore, non-reproductive adult female mice of the line selected for high litter size (S-line) and ofthe non-selected control line (C-line) were subjected to three non-social tests and asocial confrontation. We hypothesise that subjects from the S-line, which arecharacterised by a high RFI, will respond with a behavioural pattern more dominated byan active coping style when subjected to mild stressful situations, than subjects of theC-line.2. Material and methodsTwo mouse lines of the Norwegian mouse selection experiment (e.g., Vangen, 1993) wereused: a line selected for high litter size at birth (S-line) and a non-selected control line(C-line). The mice were housed in cages of 30 x 12.5 x 12.5 cm3 filled with a layer of sawdust and had free access to pellet concentrate and water. The energy content of thefood was 12.6 kJ ME per gram and contained 21% crude protein, as specified by theproducer. The mice originated from litters standardised at birth, when larger thaneight pups, to eight pups per litter. The light was left on 24 hours a day. In the 91st generation, per line, from each of 12 litters one sister-pair was randomlychosen at weaning (i.e., 3 weeks of age). Average total number of pups born was 10 and21 in the C- and S-line, respectively. All animals were housed in sister-pairs until 10weeks of age; thereafter they were housed individually. From 11-15 weeks of age thefemales were subjected to an open-field procedure, a maze, a social confrontation and arunway test.2.1. Test one: open-field testA total of 24 females of each line were introduced individually into an open-fieldapparatus at 76 and 77 days of age (OF1 and OF2). The test duration was 30 seconds.The open field (70 x 110 cm2) was surrounded by 20 cm high, non-transparent walls andwas divided into squares of 10 x 10 cm2 (Figure 1a). In OF1, the mice started in thecentre of the OF; in OF2 they started in the corner of the OF (Figure 1a). The lineswere tested in rotating order. The testing sequence of all mice was similar for bothtesting days. During the 30 seconds observation period, the position of the mouse inthe OF was registered every 5 seconds. From these data the following parameterswere measured:
    • 76 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. Vangen• Distance travelled as measured by the cumulative number of squares crossed in 5, 10, 15, 20, 25 and 30 seconds (OF1 and OF2)• Total number of squares crossed in the centre of the OF in 30 seconds (OF2) (a) 1 2 (b) G S (c) S G 25 50 100 150 200Figure 1 (a) Design of the open-field test (70 x 110 x 20 cm3). 1: starting point OF1; 2: startingpoint OF2. Dark area marks the centre of the open field. (b) Design of the maze test (55 x 55 x10 cm3). S = start box, G = goal box. Dark areas mark goal area (below G) and starting area (aboveS). Light dark areas mark dead ends. (c) Design of the runway test (200 x 10 x 10 cm3). S = startbox, G = goal box. Distance in centimetres (25 to 200) is marked.2.2. Test two: maze testThe same 24 females of each line were introduced individually into the start box of amaze at 80 to 86 days of age (M1 to M7). The test duration was 120 seconds. The maze(55 x 55 x 10 cm3) had non-transparent walls and a transparent lid on top (Figure 1b).The lines were tested in rotating order. The testing sequence of all mice was similarfor all testing days. A male stimulus mouse of the C-line (randomly selected from bothlines) was housed in the goal box of the maze at least one hour before the start of the
    • 6. Behavioural strategies in non-reproductive adult females 77test. The stimulus male was visible for the experimental female subject when the goalarea was entered. The same stimulus male was used during the whole procedure. Themaze and the start box were cleaned at the end of every session. On the sixth day ofthe procedure the maze was rotated 90° in the horizontal plane in order to confrontthe mouse with an extra-maze cue change (test M6). During M1 to M7 the followingbehavioural parameters were measured:• Latency time reaching the goal area (s) (LGA)• Latency time entering the goal box (s) (LGB)• Total time spent between reaching the goal area and entering the goal box (s) (GA-GB)• Maximum distance reached as a percentage of the total maze (as considered excluding dead-ends) (%REACHED)• Total number of encounters in dead-ends before reaching the goal area (DEAD-END)• Total number of returns to the starting area before reaching the goal area (RETURNS)2.3. Test three: social confrontation testThe same 24 females of each line were subjected to a social confrontation test at 92days of age. The test arena was a novel, unfamiliar clean cage (40 x 20 x 20 cm3). Twofemales being unfamiliar to each other and no sisters were paired for 10 minutes.Behaviour was recorded on video. Mice with different coat colours were put togetherto facilitate the distinction between the animals on video, but were further randomlyallocated to one of three interaction groups, each of which consisted of eight pairs ofmice: (1) two S-line females, (2) two C-line females, and (3) an S-line female with aC-line female. Instantaneous sampling was used to measure the occurrence ofbehavioural elements every 10 seconds. The behaviours measured are listed in Table 1.Table 1 Behavioural elements measured in the social confrontation test (after Benus, 1988)Behavioural element ExplanationFighting Behaviour shown by each of the contestants when locked together: violent kicking, biting and wrestling behaviourSubmissive upright Sitting upright, head into the air, forepaws rigidly stretched out forwardImmobility Absence of any movementSocial investigation Sniffing or nibbling any part of the opponent’s bodyJumping Jumping, often to a wallUpright Standing or sitting on hind legs, mostly making sniffing movements, with the nose up into the airSniffing Standing still with nose on the floorLocomotion Locomotion, no apparent directionGrooming Wiping, licking and nibbling the fur with forepaws and tongueMiscellaneous Any other behaviour
    • 78 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. Vangen2.4. Test four: runway testThe same 24 females of each line were introduced individually into the start box of arunway at 100 and 101 days of age (RW1 and RW2). The test duration was 60 seconds.The runway (200 x 10 x 10 cm3) had non-transparent walls and a transparent lid on top(Figure 1c). The goal box at the end of the runway was covered with saw dust. The lineswere tested in rotating order. The testing sequence of all mice was similar for bothtesting days. Latency times to reach the 25, 50, 100, 150 and 200 cm marks of therunway were recorded (Figure 1c).2.5. Data handling and statistical analysisThe SAS program was used for the statistical analysis of all traits (SAS, 1985). Linedifferences for the individual traits were tested with the model: Yij = µ + Li + eij, whereµ = overall mean, Li = effect of line i (control, selection) and eij = error term of animal jof line i, eijNID(0, σ2e). Yij denotes all traits tested with this model, all as measured onanimal j of line i: the cumulative number of squares crossed in 5, 10, 15, 20, 25 and 30seconds in OF1 and OF2, the number of lines crossed in the centre of OF2, LGA,GA-GB, %REACHED, DEAD-END and RETURNS in M1 to M7, the frequencies of‘fighting’, ‘submissive upright’, ‘immobility’, ‘social investigation’, ‘jumping’, ‘upright’,‘sniffing’, ‘locomotion’, ‘grooming’ and ‘miscellaneous’ in the social confrontation test, andthe latency times for reaching 25, 50, 100, 150 and 200 cm of RW1 and RW2. Trends and line differences in trends for cumulative number of squares crossed inOF1 and OF2 over time (5 to 30 seconds), LGA, GA-GB, %REACHED, DEAD-END, andRETURNS in the maze tests over testing day (M1 to M7, excluding M6 (extra-maze cuechange)), and latency times in RW1 and RW2 over distance reached (25 to 200 cm) aretested by fitting linear regression functions to the data. Contrasts are generated to compare cumulative number of squares crossed at 5, 10,15, 20, 25, and 30 seconds in the open-field test and for reaching 25, 50, 100, 150 and200 cm in the runway test between testing days (OF1, OF2 and RW1, RW2). The effect of the extra-maze cue change on day 6 of the maze test is tested bymeans of t-tests testing ‘H0: observed trait values = expected values as estimated fromthe regression lines’ against the alternative hypothesis that the observed trait valueson test day 6 differ from the expected values as estimated from the regression lines. Latency times of animals that do not reach the goal area in the maze tests arearbitrarily set to 121 seconds, which is 1 second higher than the full testing time.Average GA-GB is estimated only from animals that reach both GA and GB within time(i.e., within 120 seconds). Latency times for reaching, 25, 50, 100, 150 and 200 cm inthe runway tests are estimated for all animals completing the runway within time andexcluding individuals that return to the start box.
    • 6. Behavioural strategies in non-reproductive adult females 793. Results3.1. Test one: open-field testFigure 2 presents total cumulative number of squares crossed in 5 to 30 seconds in OF1and OF2. In OF1, no significant differences exist between the lines for totalcumulative number of squares crossed in 5 to 30 seconds. In OF2, S-line mice cross intotal significantly more squares in 25 and 30 seconds (P<0.05) and more squares in thecentre of the OF (8.3) than mice of the C-line (1.8; P<0.01). Females of both lines crossmore squares in OF2 than in OF1 (P<0.001). R2 values of the linear regression functions relating cumulative number of squarescrossed to time (s) are for C- and S-line females 55% and 51% in OF1 and 44% and 48%in OF2, respectively. Females of the selection line cross more squares per second (0.18and 0.31 in OF1 vs. OF2) than females of the control line (0.16 and 0.23 in OF1 vs. OF2),but this is significant in OF2 only (P<0.05). 50 OF1 60 OF2 Number of squares crossed 45 50 40 35 C 40 30 S 25 30 C 20 20 S 15 10 10 5 0 0 5 10 15 20 25 30 5 10 15 20 25 30 Time (s)Figure 2 Average total cumulative number of squares crossed in 5 to 30 seconds and standarderrors in open-field test 1 (OF1) and open-field test 2 (OF2). N = 24 per line; C = control line;S = selection line.3.2. Test two: maze testAverage LGA, GA-GB, %REACHED, DEAD-END and RETURNS in M1 to M7 are given inTable 2. S-line females reach the goal area significantly faster than C-line females inM2 to M7, but this is significantly only in M2 to M5. R2 values of the linear regressionfunctions relating LGA to testing day (M1 to M7 excluding M6 (extra-maze cuechange)) are for C- and S-line females merely 11% and 9%, respectively. Negativeregression coefficients indicate that the mice reach the goal area faster in subsequenttests (5.2 and 6.2 seconds per test for C- and S-line females, respectively; P>0.05).Females that fulfilled the maze did not always fulfil the maze or increase their latencytimes in subsequent tests.
    • 80 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. VangenTable 2 Means and standard errors (between brackets) of LGA, GA-GB, %REACHED, DEAD-ENDand RETURNS in maze test 1 to 7 (M1 to M7). M1 M2 M3 M4 M5 M6 M7 aLGA C 111.6 114.7 100.9 106.9 100.7 81.3 79.2 (3.5) (2.8) (6.4) (6.0) (7.3) (8.4) (8.5) S 114.9 91.5** 78.9* 79.4* 67.4** 78.8 75.3 (3.3) (7.4) (8.7) (8.3) (9.6) (9.5) (9.5)GA-GBb C 13.2 8.7 16.8 5.8 6.7 8.6 9.1 (4.0) (1.2) (8.2) (1.4) (0.8) (1.2) (1.7) S 8.3 5.6 4.9* 5.5 4.9 3.8** 3.0** (3.3) (0.8) (1.1) (0.9) (1.7) (0.8) (0.5)%REACHEDa C 71.1 78.2 84.1 75.8 71.3 72.1 78.5 (5.2) (3.3) (3.0) (4.7) (5.6) (6.8) (6.1) S 70.3 83.3 82.9 84.6 79.2 78.0 77.7 (3.8) (4.5) (4.9) (4.5) (6.1) (6.3) (6.3)DEAD-ENDa C 8.7 12.4 9.1 7.0 6.6 4.5 5.3 (0.8) (0.7) (0.9) (0.9) (0.8) (0.8) (0.8) S 12.3** 13.9 8.8 7.9 5.2 6.4 4.4 (0.8) (1.1) (1.1) (1.2) (0.8) (1.4) (0.8)RETURNSa C 1.0 1.5 1.1 1.3 1.2 1.6 1.0 (0.2) (0.2) (0.2) (0.3) (0.2) (0.3) (0.3) S 2.5*** 1.3 1.4 1.3 1.7 2.5 1.6 (0.4) (0.3) (0.4) (0.3) (0.5) (0.7) (0.4) a b*P<0.05; **P<0.01; ***P<0.001; C = control line; S = selection line; N = 24 per line; N from M1 toM7 = 5, 3, 6, 4, 6, 12 and 14 for the C-line, and 4, 12, 13, 14, 14, 12 and 14 for the S-line,respectively. After reaching the goal area, S-line females enter the goal box faster than C-linefemales. This is significant in M3, M6 and M7 only. The high average of the C-line in M3is due to one animal waiting 57 seconds before entering the goal box. No regressionlines could be fitted to the data on GA-GB and testing day due to too many missingvalues. No significant differences exist between the lines and no significant trend existsover subsequent maze tests for percentage reached of the total maze. Before reaching the goal area, S-line females encounter dead-ends in the mazesignificantly more often than C-line mice in M1; no significant differences between thelines exist for M2 to M7. R2 values of the linear regression functions relatingDEAD-END to testing day (M1 to M7 excluding M6 (extra-maze cue change)) are forC- and S-line females 16% and 30%, respectively. Negative regression coefficientsindicate that the females encounter fewer dead-ends in subsequent tests (0.9 and 1.6dead-ends per test for C- and S-line females, respectively); regression coefficients aresignificantly more negative in S- than in C-line females (P<0.01). Before reaching the goal area, S-line females return more often to the startingarea than C-line mice in M1; no significant line differences exist in subsequent tests.No significant trend exists over subsequent maze tests for RETURNS.
    • 6. Behavioural strategies in non-reproductive adult females 81 None of the values for DEAD-END and LGA in M6 (extra-maze cue change) differsignificantly from the expected values as predicted from the regression lines.3.3. Test three: social confrontation testAverage frequencies of ‘immobility’, ‘social investigation’, ‘upright’, ‘sniffing’, ‘locomotion’and ‘grooming’ are given in Table 3. None of the animals scored for ‘fighting’,‘submissive upright’, ‘jumping’ and ‘miscellaneous’. No significant differences existbetween the lines for ‘grooming’, ‘immobility’ and ‘upright’. S-line mice scoresignificantly higher for ‘locomotion’ and significantly lower for ‘sniffing’ and ‘socialinvestigation’.Table 3 Means and standard errors (between brackets) of frequencies of ‘immobility’, ‘socialinvestigation’, ‘upright’, ‘sniffing’, ‘locomotion’ and ‘grooming’ in the social confrontation test. immobility social upright sniffing locomotion grooming investigationC 0.0 (0.0) 8.1 (0.8) 4.5 (0.5) 13.5 (0.7) 3.9 (0.3) 0.1 (0.1)S 0.1 (0.1) 5.5 (0.5)** 5.6 (0.6) 10.5 (0.7)** 8.0 (0.7)*** 0.3 (0.1)**P<0.01; ***P<0.001; C = control line; S = selection line; N = 24 per line. 25 RW1 25.00 RW2 20 20.00 C 15 15.00 S Time (s) C 10 10.00 S 5 5.00 0 0.00 0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 Distance reached (cm)Figure 3 Average latency times (s) for reaching 25, 50, 100, 150 and 200 cm and standard errorsin runway test 1 (RW1) and 2 (RW2). N = 24 per line; C = control line; S = selection line.3.4. Test four: runway testAverage latency times (s) for reaching the 25, 50, 100, 150 and 200 cm marks in RW1and RW2 are given in Figure 3. In RW1 three, and in RW2 one S-line female areexcluded from the analyses because they returned to the start box. S-line femaleshave lower latency times at all distances in RW1 and RW2. This is significant at alldistances in RW1 (P<0.01 at 25 cm; P<0.001 at 50 to 200 cm), but only at 150 and
    • 82 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. Vangen200 cm in RW2 (P<0.05). Females of both lines are significantly faster in RW2 than inRW1 (P<0.01). R2 values of the linear regression functions relating distance reached in RW1 andRW2 to time (s) are for C- and S-line females 41% and 64% in RW1 and 64% and 49% inRW2, respectively. The intercept, a measure of time taken before entering the RW, issignificantly different from 0 in C-line females in RW1 (P<0.001), but not in S-linefemales and not in either line in RW2. In both tests, S-line females run significantlyfaster (0.05 and 0.04 s per cm in RW1 vs. RW2) than C-line females (0.08 and 0.05 sper cm in RW1 vs. RW2) (P<0.01).4. Discussion and conclusions Adult, non-reproductive female mice from a line selected for high litter size atbirth (S-line) eat more than expected from food requirements for maintenance andgrowth than mice from a non-selected control line (C-line) (Chapter 4). Since growth isvirtually absent at maturity, these differences in residual food intake (RFI) are mainlyexplained by differences in maintenance requirements. Several studies have indicatedthat a higher RFI is related to higher activity levels (e.g., Luiting et al., 1991; De Haeret al., 1993). Differences in activity suggest underlying differences in copingstrategies. For this reason, behavioural coping strategies have been investigated in non-reproductive mature C- and S-line females in four behavioural trials to investigatewhether coping strategies have been affected as a correlated effect of selection forhigh litter size.4.1 Consequences of selection for high litter size for coping behaviour in a novel environmentIn the open-field test, the total number of squares crossed is a measure of locomotionactivity. Passive copers show more immobility than active copers and will thereforeshow less locomotion. Furthermore, they will be less reluctant to enter the centre ofthe open field. In the present study, no differences in locomotion activity exist between the linesin the first open-field test. In the second open-field test, when the situation is morefamiliar and explorative behaviour is less inhibited, locomotion is increased in bothS- and C-line females. In the second open field S-line females are more active thanC-line females and are less inhibited from entering the centre of the open field. Theseresults suggest that the novelty response of S-line females is more indicative of anactive coping style than that of C-line females. In a maze test, Benus (1988) found that it took passive copers significantly moretime to reach the goal box and they made significantly less mistakes in doing so thanthe active copers. As a consequence, passive copers are expected to have longer latencytimes, reach a lower percentage of the total maze, encounter dead ends less often andreturn less often to the goal area than active copers. The ‘freezing’ reaction in passive
    • 6. Behavioural strategies in non-reproductive adult females 83copers in social interactions will prolong the time spend between reaching the goal area,from where the male mouse is visible, and entering the goal box. In the present study, S-line females show more locomotion activity in the mazetests and reach as a consequence the goal area faster than females of the C-line. Inboth lines, latency times decrease in subsequent tests, when the situation is morefamiliar. In the first maze test, when the situation is most unfamiliar, S-line femalesencounter dead ends more often and return more often to the starting area than C-linefemales. S-line females are less inhibited from entering the goal box after reachingthe goal area. The behavioural pattern displayed by S-line females in the consecutivemaze tests suggests that these mice react with a response more indicative of an activecoping style than that of C-line females. One of the most fundamental differences between active and passive copers is thedegree to which behaviour is guided by environmental stimuli (Koolhaas et al., 1997).Active copers perform very consistently over subsequent trials and show hardly anyincrease in latency times and number of errors during these runs when an extra-mazecue change is conducted. This is in contrast to passive copers, whose performances arevery variable over subsequent trials and are easily disturbed by changes in theenvironment (Benus, 1988). In the present study, no line differences are found in the reaction towards theextra-maze cue change, suggesting no differences in dependency on external stimuli.However, the differences found between aggressive and non-aggressive male mice intheir reaction to an extra-maze cue change in the study of Benus (1988) were found inmice which all reached the goal box errorless within 15 seconds before the extra-mazecue change was conducted. This criterion was not fulfilled in the present study. Coping behaviour in social encounters is mainly described in male mice. Sinceaggressive behaviour is mediated through testosterone levels, females are notexpected to display the same behaviour in a social confrontation. When fighting occurs,active copers are characterised by ‘fight’ or ‘flight’ behaviour, whereas passive coperswill exhibit more ‘immobility’ and ‘submissive upright’ (Benus, 1988). Coping behaviourwhen fighting is absent has not been described clearly in the literature. Passive copersare expected to display more ‘sniffing’, ‘upright’ and ‘social investigation’ in order togain information about the environment. Active copers are expected to show more‘locomotion’ activity. ‘Jumping’ may be considered as a way to escape from the stressorand thus more ‘typical’ for active copers. In the present study, in the social confrontation test, S-line females showed morelocomotion activity than C-line females. C-line females were more immobile andinvestigated the floor and opponent more than S-line females. These results suggestthat the behaviour of S-line females in a social confrontation is more related to anactive coping style than the behaviour of C-line females. However, since no fightingoccurred between C- and S-line females in the present study, this may not be a verygood test to distinguish between female active and passive copers. The runway test measures the locomotion activity, or speed, in a situation where themice are undisturbed by dead ends. Active copers will show more locomotion activityand will therefore be faster than passive copers. Moreover, they will be less reluctantto enter the runway and consequently have shorter latency times.
    • 84 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. Vangen In the present study, in the first runway test, the positive intercept of the linearrelationship between distance reached and time in C-line females indicates that theseanimals took some time before entering the runway. The intercept in S-line females isnot significantly different from zero, which indicates that these animals startedrunning practically immediately. Furthermore, S-line females ran faster in both runwaytests than C-line females. These results show that S-line females show behaviour thatis more indicative of an active coping strategy than C-line females. Summarised, the results of the present study suggest that long-term selection forhigh litter size has to a certain extent affected individual coping strategies in adultnon-reproductive female mice as a correlated response. The behavioural differences asmeasured in the four behavioural trials point in the same direction: the results suggestthat the novelty response of S-line females is more dominated by an active coping stylethan that of C-line females. However, more tests, such as additional behavioural tests(e.g., measurement of circadian rhythmicity of activity), physiological tests(measurement of stress-response in blood pressure, heart beat, and corticosteroneconcentrations), and neuro-biological tests (e.g., measurement of apomorphine inducedstereotypies) are required to be more conclusive on whether selection for high littersize has resulted in mice that adopt the active coping strategy.4.2 A biological explanation to the relationship between residual food intake and activity levelsThe results support the observation that animals with highest RFI have highest activitylevels. This relationship can be explained by the Resource Allocation Theory developedby Beilharz et al. (1993). Resources consumed by physiological processes (e.g.,maintenance, (re)production, physical activity, reaction to pathogens and stressors) addto give the total amount of resources consumed (e.g., food intake, body tissuemobilisation), since resources consumed by one process can not be allocated to anotherprocess:R = (kA x A) + (kB x B) + Σ(kQi x Qi),where R = total amount of available resources, k = resource conversion factor,(kA x A) = resources used for maintenance, (kB x B) = resources used for production andΣ(kQi x Qi) = resources used for processes other than maintenance and production. Thisequation shows a remarkable similarity with the equation for the calculation of RFI(according to Luiting and Urff, 1991):FC = (b1 x BW0.75) + (b2 x BWG) + e,where FC = food consumption, BW0.75 = metabolic body weight, BWG = body weight gain,b1 and b2 = partial regression coefficients representing maintenance requirements andrequirements for production, respectively, and e = residual term representing RFI.When animals are (re)producing, traits such as egg weight, milk volume, litter size andlitter weight should be included as covariates in the model.
    • 6. Behavioural strategies in non-reproductive adult females 85 This similarity implies that RFI quantifies the total amount of resources available toan animal for other processes than maintenance and production: animals with high RFIhave more ‘buffer’ resources left for, e.g., physical activity and the ability to cope withunexpected stresses (Luiting et al., 1997). Indeed, laying hens with low RFI were lessadapted to cope with high temperatures (Bordas and Minvielle, 1997) and had a weakercorticosteroid response but maintained elevated corticosterone levels for a longer timeafter injection with ACTH (Luiting et al., 1994). Consequently, RFI can be used as a toolto deal with welfare matters. Animals with a very low RFI will be more susceptible tostress. Since RFI gives an estimate of available buffer resources relative to thecontrol population on which the model is formed, an appropriate control population hasto be defined. RFI can be used to quantify differences in buffer resources between,e.g., breeds, lines and housing systems. Estimation of RFI is a practical tool since inmost species body weight, food intake and output traits such as eggs and milk (weight,number, volume) are traits that are relatively easy to measure. Chapter 4 showed that adult non-reproductive males of the S-line have a higherRFI than males of the C-line. Adult males of the S-line show generally more activity inthe maze test and the social interaction test, although mostly non-significant(unpublished results). These results suggest that with selection on a trait performedby females only, males are influenced to a lesser extent or only in the longer term. It is suggested that the fact that particularly females of the S-line increase theirRFI during the adult period, compared with C-line mice and males of the S-line, may bean anticipation to the metabolically stressful and energy demanding periods pregnancyand lactation to support a genetically highly increased litter size (Chapter 4). If thisincrease in RFI is not sufficient, available metabolic resources may drop considerablyand the selected animals may be left with inadequate resources to respond to stressfactors. Future research will investigate the resource situation and its consequencesfor behavioural strategies in lactating females.AcknowledgementsThis study has been supported by a grant from the Norwegian Research Council,project number 114258/111, ‘Consequences of selection for high litter size for theability to cope in a stressful environment’. Kari Kjus is gratefully acknowledged forcarrying out the Norwegian mouse selection experiment and her help in providing andmaintaining the mice of the present study. Pieter Knap is most gratefully thanked forhis useful comments on the manuscript.
    • 86 W.M. Rauw, P. Luiting, M. Bakken, C.J.M. de Veer, T. Schuurman and O. VangenReferences• Beilharz, R.G., Luxford, B.G., Wilkinson, J.L., 1993. Quantitative genetics and evolution: Is our understanding of genetics sufficient to explain evolution? J. Anim. Breed. Genet. 110:161-170.• Benus, R.F., 1988. Aggression and Coping. Differences in behavioural strategies between aggressive and non-aggressive male mice. PhD Thesis Rijksuniversiteit Groningen, The Netherlands.• Benus, R.F. and Röndigs, M., 1996. Patterns of maternal effort in mouse lines bidirectionally selected for aggression. Anim. Behav. 51:67-75.• Bordas, A. and Minvielle, F., 1997. Réponse à la chaleur de poules pondeuses issues de lignées sélectionnées pour une faible (R-) ou forte (R+) consommation alimentaire résiduelle. Genet. Sel. Evol. 29:279-290.• Braastad, B.O. and Katle, J., 1989. Behavioural differences between laying hen populations selected for high and low efficiency of food utilisation. Br. Poultry Sci. 30:533-544.• Compaan, J.C., van Wattum, G., de Ruiter, A.J.H., van Oortmerssen, G.A., Koolhaas, J.M., Bohus, B., 1993. Genetic-differences in female house mice in aggressive response to sex steroid-hormone treatment. Phys. Beh. 54:899-902.• Cools, A.R., Brachten, B.R., Heeren, D., Willemen, A., Ellenbroek, B., 1990. Search after neurobiological profile of individual-specific features of Wistar rats. Brain Research Bulletin 24:49-69.• De Haer, L.C.M., Luiting, P., Aarts, H.L.M., 1993. Relations among individual (residual) feed intake, growth performance and feed intake pattern of growing pigs in group housing. Livest. Prod. Sci. 36:233-253.• Driscoll, P., Dedek, J., D’Angio, M., Claustre, Y., Scatton, B., 1990. A genetically-based model for divergent stress responses: behavioral, neurochemical and hormonal aspects. Adv. Anim. Breed. Genet. 5:97-107.• Hessing, M.J.C., Hagelsø, A.M., Van Beek, J.A.M., Wiepkema, P.R., Schouten, W.G.P., Krukow, R., 1993. Individual behavioural characteristics in pigs. Appl. Anim. Behav. Sci. 37:285-295.• Hopster, H., 1998. Coping strategies in dairy cows. PhD-thesis, Agricultural University of Wageningen, Wageningen, The Netherlands.• Koolhaas, J.M., de Boer, S.F., Bohus, B., 1997. Motivational systems or motivational states: behavioural and physiological evidence. Appl. Anim. Behav. Sci. 53:131-143.• Luiting, P., 1990. Genetic variation of energy partitioning in laying hens: causes of variation in residual feed consumption. World’s Poultry Sci. J. 46:133-152.• Luiting, P., Decuypere, E., De Groot, P.N., Buyse, J., Room, G., 1994. Selection for feed efficiency and consequences for stress susceptibility. 45th Meeting of the EAAP, Edinburgh, United Kingdom.• Luiting, P., Schrama, J.W., Van der Hel, W., Urff, E.M., 1991. Metabolic differences between White Leghorns selected for high and low residual feed consumption. Br. Poultry Sci. 32:763- 782.• Luiting, P. and Urff, E.M., 1991. Optimization of a model to estimate residual feed consumption in the laying hen. Livest. Prod. Sci. 27:321-338.• Luiting, P., Vangen, O., Rauw, W.M., Knap, P.W., Beilharz, R.G., 1997. Physiological consequences of selection for growth. 48th Annual Meeting of the EAAP, Vienna, Austria.• Morrison, W.D. and Leeson, S., 1978. Relationship of feed efficiency to carcass composition and metabolic rate in laying birds. Poultry Sci. 57:735-739.• SAS, 1985. SAS user’s guide: statistics. Version 5. Sas Institute Inc., Cary, NC, USA.
    • 6. Behavioural strategies in non-reproductive adult females 87• Sluyter, F., Bult, A., Lynch, C.B., van Oortmerssen, G.A., Koolhaas, J.M., 1995. A comparison between house mouse lines selected for attack latency or nest-building: evidence for a genetic basis of alternative behavioral strategies. Behav. Gen. 25:247-252.• Sluyter, F., van Oortmerssen, G.A., de Ruiter, A.J.H., Koolhaas, J.M., 1996. Aggression in wild house mice: current state of affairs. Behav. Genet. 26:489-496.• Vangen, O., 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.• Verbeek, M.E.M., Boon, A., Drent, P.J., 1996. Exploration, aggressive behaviour and dominance in pair wise confrontations of juvenile male great tits. Behav. 133:945-963.• Von Felde, A., Roehe, R., Looft, H., Kalm, E., 1996. Genetic association between feed intake and feed intake behaviour at different stages of growth of group-housed boars. Livest. Prod. Sci. 47:11-22.• Wechsler, B., 1995. Coping and coping strategies: a behavioural view. Appl. Anim. Behav. Sci. 43:123-134.
    • 7 FOOD RESOURCE ALLOCATION PATTERNS IN LACTATING FEMALES W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. LuitingAbstract In an earlier study it was shown that mature non-reproductive individuals(6 to 10 weeks of age) from a mouse line selected for more than 90 generations forhigh litter size at birth (S-line) have a significantly higher residual food intake (RFI)than mice of a non-selected control line (C-line), in particular females of this line. Itwas suggested that this higher RFI in non-reproductive females may anticipate thehighly increased resource demand during pregnancy and especially lactation. For thisreason, RFI was investigated in both lines in lactating females with litters that werestandardised at birth (Ss and Cs) and in lactating females with non-standardised litters(Sns and Cns). Residual food intake during lactation was significantly lower in Sns-damsthan in Cns-line dams. After weaning Sns-dams seem to be able to restore the negativeresource situation. Residual food intake during lactation was lower in Sns-dams than inSs-dams. After weaning, no differences were found between dams with formerlystandardised and non-standardised litters. Pre-weaning mortality rate was higher inS-line non-standardised litters than in C-line non-standardised litters. Degree ofmaturity was higher in C-line pups of non-standardised litters than in S-line pups ofnon-standardised litters and higher in pups of standardised litters than in pups of non-standardised litters. Results of the present study suggest that dams selected for highlitter size allocated considerably more resources to maintenance of offspring than non-selected dams, which may put them higher at risk to behavioural, physiological andimmunological problems. This was insufficient to provide the offspring with an adequateamount of resources, resulting in reduced pup development and increased pre-weaningmortality rates.Keywords: mice, lactation, resource allocation, residual food intakeBased on W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting, Differences in food resourceallocation patterns in lactating females in a long-term selection experiment for litter size in mice,submitted.
    • 90 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting1. IntroductionChapter 4 showed that mature non-reproductive individuals (6 to 10 weeks of age) froma mouse line selected for more than 90 generations for high litter size at birth (S-line),and in particular females, have a significantly higher residual food intake (RFI) thanmice of a non-selected control line (C-line). Residual food intake is defined as the part of the food intake that is unaccountedfor by food requirements for maintenance and production, or in other words, as thedifference between the food that is consumed by an animal and its consumption aspredicted from a model involving its growth and maintenance requirements. Variation inRFI can be caused by variation in partial efficiencies for maintenance and growth andby variation in metabolic food demanding processes not included in the model, such asbehavioural activities, responses to pathogens and responses to stress. Since growth isvirtually absent at maturity, the differences in RFI are mainly explained by differencesin maintenance requirements (Luiting, 1990). It is interesting that particularly females of the selection line have a very high RFI,since these animals can express the trait their genotype has been selected for: a highlitter size at farrowing. This higher RFI in non-reproductive females may anticipate thehighly increased resource demand during pregnancy and especially lactation. However,these periods can be expected to considerably change the physiological resourcedemand, since the selection-line females have to support a genetically highly increasedlitter size. The question is whether this can be supported by an increase in food intakeduring these periods, or whether the RFI will drop considerably when reproductiveperformance is included in its calculation. In the present study we investigated residual food intake and offspring developmentfrom birth to weaning in a long-term selection experiment for litter size in mice, inlactating females with litters that were standardised at birth and in lactating femaleswith non-standardised litters. The aim was to study the food resource allocationpatterns in these animals and the consequences for offspring development.2. Materials and methodsTwo mouse lines of the Norwegian mouse selection experiment (e.g., Vangen, 1993) wereused: a line selected for 104 generations for high litter size at birth (S-line) and a non-selected control line (C-line). Average total number of pups born in the 104th generationwas 10 in the C-line and 21 in the S-line. Per line, 98 females were randomly chosen at 3 weeks of age (i.e., at weaning) andhoused individually. The mice originated from litters standardised at birth, when largerthan eight pups, to eight pups per litter. At 10 weeks of age all females were mated andstayed with the male for 2 weeks. From 87 C-line females and 96 S-line females thatbecame pregnant, the litters of 45 C-line dams (Cs) and 48 S-line dams (Ss) werestandardised at birth, when larger than 8 pups, to 8 pups per litter; the litters of 42C-line dams (Cns) and 48 S-line dams (Sns) were not standardised. During the periodfrom farrowing to weaning all pups of 2 Cns-litters, 6 Cs-litters, 1 Sns-litter and 1Ss-litter, and 1 Cs-line dam died.
    • 7. Food resource allocation patterns in lactating females 91 At 13 and 15 days in lactation, 20 Cns-dams, 20 Cs-dams, 20 Sns-dams and 20Ss-dams were subjected to an open-field test and a runway test (test duration of 60seconds), as described in Chapter 6. Since the calculations of residual food intake werenot affected by these tests, these animals were included in the analysis of the presentstudy. The mice were housed in 30 x 12.5 x 12.5 cm3 cages bedded with sawdust and hadfree access to pellet concentrate and water. The energy content of the food was12.6 kJ ME per gram and contained 21% crude protein, as given by the producer (similarto Chapter 3). Light was left on 24 hours a day.2.1. Non-reproductive females2.1.1. Body weight and food intakeFrom 21 to 69 days of age, individual body weight (g) and food intake (g/3d) weremeasured every three days. From this data individual body weight gain (g/3d) andcumulative food intake (g) were calculated.2.1.2. Residual food intakeResidual food intake (g/3d) was estimated, according to Luiting and Urff (1991), frommultiple linear regression of food intake (g/3d) on metabolic body weight (kg0.75) andbody weight gain (g/3d). Residual food intake is defined as the difference between thefood that is consumed by an animal and its consumption as predicted from requirementsfor growth and maintenance per metabolic kg of the C-line female population. Themethod for estimating RFI in non-reproductive mice is extensively described in Chapter4. Residual food intake was estimated for each 3-day period from 21 to 69 days of ageand was subsequently expressed on a daily basis (g/d). Residual food intake wasfurthermore estimated for a ‘growing period’, i.e., from 21 to 42 days of age, and an‘adult period’, i.e., from 42 to 69 days of age (Chapter 4). Residual food intake wasestimated from accumulated data on growth and food intake per animal over theseperiods. Metabolic body weight of the growing period was estimated as the average ofmetabolic body weights for all 3-day periods from 21 to 42 days of age. Metabolic bodyweight of the adult period was estimated as the average of metabolic body weights forall 3-day periods from 42 to 69 days of age. Residual food intake for the growingperiod and the adult period were subsequently expressed on a daily basis (g/d). Residual food intake for each animal in the C-line was obtained as the error term ofthe linear regression. This implies that the average RFI of all C-line females equalled 0(Chapter 4).2.1.3. Asymptotic mature body weight and mature food intakeFollowing Archer and Pitchford (1996), modified Parks’ (1982) curves were fitted toindividual data on body weight (g) against cumulative food intake (g) from 21 to 69 days
    • 92 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luitingof age, yielding, among other parameters, individual estimates of asymptotic mature(virgin) body weight (A in g). A linear function by Parks (1982, p. 31) was fitted torelate individual data on cumulative food intake (g) to age (d), yielding individualestimates of mature (virgin) daily food intake (MFI in g/d). The methods for theestimation of A and MFI are extensively described in Chapter 4.2.2. Lactating females2.2.1. Body weight, food intake and litter traitsFrom farrowing to weaning (i.e., three weeks in lactation), per family (i.e., dam + litter),maternal body weight (g), litter weight (g), litter size and food intake (g/d) weremeasured daily. From these data, per family, pup weight (i.e., litter weight divided bylitter size) (g), maternal body weight gain (g/d), pup body weight gain (g/d) andcumulative food intake (g) were calculated. Furthermore, per family, the day that thepups opened their eyes was recorded. Pre-weaning mortality rate in families with non-standardised litters was calculatedas ‘total number of pups that died from birth to weaning’ expressed as a percentage of‘total number of pups born’. Pre-weaning mortality rate in families with standardisedlitters was calculated as ‘total number of pups that died from birth to weaning afterstandardisation’ expressed as a percentage of ‘number of pups after standardisation’. The moment of peak lactation (days in lactation) was estimated by fitting a quadraticfunction to data on maternal body weight (g) against days in lactation. The R2 value of aquadratic regression of milk yield on age was 92.3% in the study of Jara-Almonte andWhite (1972). In the present study it is assumed that the increase and subsequentdecrease in body weight of the dam during lactation results from an increase andsubsequent decrease in milk production:BW = a + bD + cD2, (1)where BW = body weight of the dam (g) and D = days in lactation; a, b, c are parametersto be estimated. Days in lactation at peak lactation was estimated by solving the firstderivative of this function for:DP = -b/2c,where DP = days in lactation at peak lactation, and b and c are the parametersdescribed in (1). For each individual family, relative increase in maternal body weight during lactationcompared with mature virgin body weight was calculated as maternal body weight (g)divided by the individual estimate of asymptotic mature virgin body weight (A in g)multiplied by 100%. Litter weight during lactation relative to mature virgin maternalbody weight was calculated as litter weight (g) divided by the individual estimate of A(g) of the dam multiplied by 100%. Degree of maturity of pups was calculated, accordingto Taylor and Murray (1987), as pup body weight (g) divided by the individual estimate
    • 7. Food resource allocation patterns in lactating females 93of A (g) of the dam multiplied by 100% (degree of maturity is calculated as body weightdivided by mature body weight of the animal but since no data were available toestimate individual mature body weight of the offspring, instead, the estimate ofasymptotic mature virgin body weight of the dam was used as a scaling factor). Relativeincrease in food intake during lactation compared with mature virgin maternal foodintake was calculated as food intake (g/d) divided by the individual estimate of maturevirgin food intake (MFI in g/d) multiplied by 100%.2.2.2. Residual food intakeThe equation used to estimate RFI (g/d) for each family was based on the followingmultiple linear regression of food intake (g/d) on maternal metabolic body weight(kg0.75), maternal body weight gain (g/d), litter size, pup metabolic body weight (g),litter metabolic weight (i.e., pup metabolic body weight multiplied by litter size; in g),pup body weight gain (g/d) and litter weight gain (i.e., pup body weight gain multipliedby litter size; in g/d) in control-line families with non-standardised litters (Cns):FCi(Cns) = b0(Cns) + (b1(Cns) x DBW0.75i(Cns)) + (b2(Cns) x DBWGi(Cns)) + (b3(Cns) x LSi(Cns)) + (b4(Cns) x PBWi(Cns)) + (b5(Cns) x (LSi(Cns) x PBWi(Cns))) + (b6(Cns) x PBWGi(Cns)) + (b7(Cns) x (LSi(Cns) x PBWGi(Cns))) + ei(Cns),which can be written as:FCi(Cns) = b0(Cns) + (b1(Cns) x DBW0.75i(Cns)) + (b2(Cns) x DBWGi(Cns)) + (b3(Cns) x LSi(Cns)) + ((b4(Cns) + b5(Cns) x LSi(Cns)) x PBWi(Cns)) + ((b6(Cns) + b7(Cns) x LSi(Cns)) x PBWGi(Cns)) + ei(Cns), (2)where FCi(Cns) = food consumption of Cns-family i (g/d); DBW0.75i(Cns) = metabolic bodyweight of the dam of Cns-family i (kg0.75); DBWGi(Cns) = body weight gain of the dam ofCns-family i (g/d); LSi(Cns) = litter size of Cns-family i; PBWi(Cns) = average metabolic bodyweight of a pup of Cns-family i (g); PBWGi(Cns) = average body weight gain of a pup ofCns-family i (g/d); b0(Cns) = Cns-line population intercept; b1(Cns), b2(Cns), b3(Cns), b4(Cns), b5(Cns),b6(Cns), b7(Cns), = Cns-line population partial regression coefficients and ei(Cns), = the errorterm, representing RFI of Cns-family i (g/d). b1(Cns) and b2(Cns) represent maintenancerequirements per metabolic kg and requirements for growth of the dam, respectively;b3(Cns) extrapolates food requirements per average pup to food requirements per litter;(b4(Cns) + b5(Cns) x LSi(Cns)) represents litter-size-independent maintenance requirementsper metabolic g of a pup; (b6(Cns) + b7(Cns) x LSi(Cns)) represents litter-size-independentfood requirements for growth of a pup. Equation (2) was fitted per day from farrowingto 3 weeks in lactation. Subsequently, RFI of C-line families with standardised litters (Cs) and all S-linefamilies (Sns and Ss) was estimated as:
    • 94 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting ˆ ˆRFIi(Cs,Sns,Ss) = FCi(Cs,Sns,Ss) - { b 0(Cns) + ( b 1(Cns) × DBW0.75i(Cs,Sns,Ss)) + ˆ ˆ ( b 2(Cns) x DBWG i(Cs,Sns,Ss)) + ( b 3(Cns) x LS i(Cs,Sns,Ss)) + ˆ ˆ (( b 4(Cns) + b 5(Cns) x LS i(Cs,Sns,Ss)) x PBW i(Cs,Sns,Ss)) + ˆ ˆ (( b 6(Cns) + b 7(Cns) x LS i(Cs,Sns,Ss)) x PBWG i(Cs,Sns,Ss))}, (3)where RFIi(Cs,Sns,Ss) = residual food intake of Cs-, Sns- and Ss-family i (g/d); FCi(Cs,Sns,Ss) =food consumption of Cs-, Sns- and Ss-family i (g/d); DBW0.75i(Cs,Sns,Ss) = metabolic bodyweight of the dam of Cs-, Sns- and Ss-family i (kg0.75); DBWGi(Cs,Sns,Ss) = body weightgain of the dam of Cs-, Sns- and Ss-family i (g/d); LSi(Cs,Sns,Ss) = litter size of Cs-, Sns-and Ss-family i; PBWi(Cs,Sns,Ss) = average metabolic body weight of a pup of Cs-, Sns- andSs-family i (g); and PBWGi(Cs,Sns,Ss) = average body weight gain of a pup of Cs-, Sns- and ˆ ˆSs-family i (g/d); b 0(Cns) to b 7(Cns) are the estimates of b0(Cns) to b7(Cns) described in (2).This was done using the estimates from each day of measurements from farrowing tothree weeks in lactation. Respiration rate (R) as a function of body mass (BW) can usually be expressed bymeans of the equation R = aBWb. The b-value is often close to ¾ when organismscovering a large span in body mass are compared. Riisgård (1998) concluded that the‘¾ power scaling law’ is a formalism and not a biological law because respiration andgrowth are integrated through the energetic costs of growth. Young and fast growingstages show usually higher weight specific respiration rates (b ~ 1) than older and adultstages (Riisgård, 1998). Because the exact weight specific respiration rates are notknown in the present study, metabolic body weight of a pup from birth to weaning isestimated as BW1, whereas metabolic body weight of mice after 3 weeks of age isestimated as BW0.75. The experimental period was subsequently divided into a period from farrowing topeak lactation (i.e., from 0 to 2 weeks in lactation; F-PL), and a period from peaklactation to weaning (i.e., from 2 to 3 weeks in lactation; PL-W). Hammond and Diamond(1992) and Millican et al. (1987) define peak lactation as the 15th day after parturition.Hanrahan and Eisen (1970) and Jara-Almonte and White (1972) observed that milk yieldin mice peaked at about 13 days in lactation. In the present study we defined peaklactation as the 14th day after parturition. Equation (2) was fitted for the F-PL periodand PL-W period from accumulated data on growth and food intake per family overthese periods. Maternal and pup metabolic body weights of the F-PL period wereestimated as the average of the daily metabolic body weights from farrowing to peaklactation. Maternal and pup metabolic body weights of the PL-W period were estimatedas the average of the daily metabolic body weights from peak lactation to weaning. Residual food intake for each Cns-family was obtained as the error term ei(Cns) in (2)and RFI of Cs, Sns and Ss families was estimated as (3). This implies that the averageRFI of all Cns-families equalled 0. Residual food intake was consequently expressed on adaily basis, for the F-PL period and for the PL-W period.
    • 7. Food resource allocation patterns in lactating females 952.3. After weaning2.3.1. Body weight and food intakeFor each dam, from weaning to 25 days after weaning the offspring, individual bodyweight (g) and food consumption (g/5d) were measured every five days. From thesedata individual body weight gain (g/5d) and cumulative food intake (g) were calculated.2.3.2. Residual food intakeResidual food intake (g/5d) was estimated as in section 2.1. Residual food intake wasestimated for each 5-day period from weaning to 25 days after weaning and wassubsequently expressed on a daily basis (g/d). Residual food intake was subsequently estimated for the total ‘after weaning’ periodfrom weaning to 25 days after weaning from accumulated data on growth and foodintake over this period. Metabolic body weight of the female was estimated as theaverage of metabolic body weights for all 5-day periods from weaning to 25 days afterweaning. Residual food intake was subsequently expressed on a daily basis (g/d). Residual food intake for each female of the C-line was obtained as the error term ofthe linear regression. This implies that the average RFI of all C-line females equalled 0(Chapter 4).2.4. Data analysisThe SAS program was used for the statistical analysis of all traits (Statistical AnalysisSystem Institute, 1985). Line differences for the individual traits were tested withthe model:Yij = µ + Li + eij,where µ = overall mean, Li = effect of line i (control, selection) and eij = error term ofanimal j of line i, eijNID(0,σ2e). Yij denotes all traits tested with this model, all asmeasured on animal j of line i: RFI for each 3-day period from 21 to 69 days of age,RFI in the growing period, RFI in the adult period, A and MFI in non-reproductivefemales; number of liveborn pups, number of stillborn pups and pre-weaning mortalityrate in lactating females; RFI for each 5-day period from weaning to 25 days afterweaning and RFI in the ‘after weaning’ period in dams after weaning. Pre-weaningmortality rate was tested for line effect within each standardisation level. Differences between lines and levels of standardisation for the individual traitswere tested with the model:Yijk = µ + Li + Sj + (LS)ij + eijk,where µ = overall mean, Li = effect of line i (control, selection), Sj = effect ofstandardisation j (non-standardised, standardised), (LS)ij = interaction effect of line i
    • 96 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luitingwith standardisation j, and eijk = error term of animal k of line i and standardisation j,eijkNID(0,σ2e). Yijk denotes all traits tested with this model, all as measured on animal kof line i and standardisation j: daily maternal body weight, litter weight, pup weight,food intake, maternal body weight relative to A, litter weight relative to A, pup weightrelative to A, food intake relative to MFI, litter size at weaning, the day that the pupsopen their eyes, and days in lactation at peak lactation in lactating females and bodyweight and food intake for each 5-day period from weaning to 25 days after weaning indams after weaning the offspring.3. Results3.1. Non-reproductive femalesAverage body weight and food intake in non-reproductive males and females from 3 to10 weeks of age in the 92nd and 95th generations of the C- and S-line have beenextensively described in Chapter 3. The present study (females only) gave similarresults.3.1.1. Residual food intakeFigure 1 shows for each line the average RFI for each 3-day period from 21 to 69 daysof age. R2 values of the multiple regressions per 3-day period were in the range of 21%to 69%. From 24 to 30 days of age, S-line females had a significantly lower RFI thanC-line females (P<0.001). From 33 to 69 days of age, S-line females had a higher RFIthan C-line females; this was significant from 42 to 69 days of age (P<0.001). Figure 1 shows that RFI from 21 to 69 days of age in C-line females was 0. Thereason for this is that the equation used to estimate RFI was based on all C-linefemales and hence the average RFI of the C-line population was 0. Figure 1 shows anincreasing trend in S-line females: RFI of the S-line was lower than RFI of the C-lineduring the fourth week of age and higher from the sixth to the tenth weeks of age. Average RFI per line in the growing period and the adult period are presented inFigure 2. R2 values of the multiple regressions per period were 74% for the growingperiod and 29% for the adult period. Residual food intake during the growing periodwas not significantly different between the lines; in the adult period, S-line femaleshad a significantly higher RFI than C-line females (P<0.001).3.1.2. Asymptotic mature body weight and mature food intakeR2 values of Parks (1982) growth curves, relating body weight to cumulative food intake,were in the range of 80% to nearly 100%; R2 values of individual linear regressions,relating cumulative food intake to age, were all nearly 100%. Estimates (± standarderror) of mature body weight (A in g) were 28.8 ± 0.249 for C-line females and38.7 ± 0.367 for S-line females. ‘A’ was significantly higher in the S-line than in theC-line (P<0.001). Estimates of mature food intake (MFI in g/d) were 4.66 ± 0.0306 for
    • 6 After Growing Adult F-PL PL-W 5 period period weaning 4 3 2 1 0 -1 Residual food intake (g/d) -2 -3 -4 -5 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 5 10 15 20 25 Age (d) Days in lactation Days after weaning 7. Food resource allocation patterns in lactating females C S Cns Cs Sns SsFigure 1 Average daily residual food intake (g/d) during the ‘growing period’, the ‘adult period’, from farrowing to peak lactation (F-PL), from peaklactation to weaning (PL-W) and ‘after weaning’. C = control line; S = selection line; ns = with non-standardised litters; s = with standardised litters. 97
    • 98 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. LuitingC-line females and 6.14 ± 0.0480 for S-line females. MFI was significantly higher in theS-line than in the C-line (P<0.01). Growing Adult F-PL PL-W After 1.2 period period weaning 1.0 0.8 Residual food intake (g/d) 0.6 0.4 0.2 0.0 -0.2 C S C S Cns Cs Sns Ss Cns Cs Sns Ss C S -0.4 -0.6 -0.8 -1.0Figure 2 Average residual food intake (g/d) during the ‘growing period’, the ‘adult period’, fromfarrowing to peak lactation (F-PL), from peak lactation to weaning (PL-W) and ‘after weaning’.C = control line; S = selection line; ns = with non-standardised litters; s = with standardisedlitters.3.2. Lactating females3.2.1. Body weight, food intake and litter traitsTable 1 presents, per line, average number of liveborn pups and average number ofstillborn pups. Table 1 shows furthermore for each standardisation level in each lineaverage litter size at weaning, average pre-weaning mortality rate and the average daythat the pups open their eyes. Pre-weaning mortality rate represents for non-standardised litters the percentage of pups that died from birth to weaning (includingstillborn pups) and for standardised litters the percentage of pups that died frombirth to weaning after standardisation. Therefore, pre-weaning mortality rate wastested for a line effect within standardisation level. Number of liveborn pups was about twice as high in the S-line as in the C-line.Number of stillborn pups was significantly higher in the S-line than in the C-line.Pre-weaning mortality rate in non-standardised litters was significantly higher in theS-line than in the C-line; pre-weaning mortality rate in standardised litters wassignificantly higher in the C-line than in the S-line. C-line pups opened their eyes earlierthan S-line pups and pups from standardised litters opened their eyes earlier than pupsfrom non-standardised litters (Table 1). Figures 3a to 3d present for each standardisation level in each line averagematernal body weight (Figure 3a), average litter weight (Figure 3b), average pup bodyweight (Figure 3c) and average food intake (Figure 3d) from farrowing to weaning.
    • 7. Food resource allocation patterns in lactating females 99Table 1 Means and standard errors of number of liveborn pups and number of stillborn pups, perline, and litter size at weaning, pre-weaning mortality rate, the day that the pups open their eyes,and days in lactation (d in lact) at peak lactation, for each standardisation level in each line. C-line S-line Cns Cs Sns SsNumber liveborn pups 10.3 ± 0.262 20.2*** ± 0.327Number stillborn pups 0.287 ± 0.0748 0.667* ± 0.135 a b cLitter size at weaning 8.52 ± 0.475 6.00 ± 0.377 13.5 ± 0.484 7.60a ± 0.178Pre-weaning mortality (%) 18.1 ± 3.94 21.3 ± 4.83 35.6 ± 2.76***1 4.95 ± 2.23**2 a b cEyes open (d in lact) 13.2 ± 0.0940 12.8 ± 0.0961 13.9 ± 0.0782 13.2a ± 0.0784Peak lactation (d in lact) 15.8a ± 1.49 12.1b ± 0.957 14.6ac ± 0.524 13.2bc ± 0.273Within a row, means without a common superscript letter differ (P<0.05). 1Sns compared withCns; 2Ss compared with Cs; *P<0.05; **P<0.01; ***P<0.001; C = control line; S = selection line;ns = with non-standardised litters; s = with standardised litters. From farrowing to weaning, S-line dams were significantly heavier than C-line dams(P<0.001). Dams with non-standardised litters were heavier than dams withstandardised litters, but this is significant at 4, 7 to 9, 12 to 14, and 18 to 21 days inlactation only (P<0.05) (Figure 3a). From birth to weaning, S-line litters were heavier than C-line litters (P<0.001).Non-standardised litters were heavier than standardised litters, but in the C-line thiswas significant from birth to 2 days in lactation, from 8 to 12 and at 21 days inlactation only (P<0.05) (Figure 3b). At birth, average pup weight was similar for each line and each standardisationlevel. From 1 to 21 days in lactation, pups of Ss-families were heavier than pups of Sns-,Cns- and Cs-families (P<0.001). From 2 to 21 days in lactation, pups of Cs-families wereheavier than pups of Cns-families (P<0.01) and from 3 to 20 days in lactation, pups ofCs-families were heavier than pups of Sns-families (P<0.05). Pups of Sns-families wereheavier than pups of Cns-families at 21 days in lactation only (P<0.01) (Figure 3c). Food intake is considerably increased during lactation. From farrowing to weaning,S-line families ate more than C-line families (P<0.001). Families with non-standardisedlitters ate more than families with standardised litters; in the C-line this wassignificant at 9, and 18 to 21 days in lactation only (P<0.05) and in the S-line this wassignificant at 3 to 10 and 19 to 21 days in lactation (P<0.05) (Figure 3d). The estimates of days in lactation at peak lactation are presented in Table 1. 2R values of the quadratic functions were in the range of merely 14% to 93%. Peaklactation in dams with non-standardised litters was later than in dams withstandardised litters, but this was significant in C-line families only. Figures 3e to 3h present for each standardisation level in each line, from farrowingto weaning, average maternal body weight relative to asymptotic mature virgin bodyweight (A) (Figure 3e), average litter weight relative to A (Figure 3f), average pup bodyweight relative to A (Figure 3g), and average food intake relative to mature virgin foodintake (MFI; Figure 3h). From 5 days in lactation to weaning, S-line dams were significantly heavier relativeto A than C-line dams (P<0.05). Dams with non-standardised litters were significantly
    • 100 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting a e 65 161 (Body weight/A) x 100% Body weight (g) 60 55 151 50 141 45 40 131 35 30 121 f b 175 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 420 (Litter weight/A) x 100% 150 340 Litter weight (g) 125 100 260 75 180 50 100 25 0 20 c 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 g 20 50 (Pup weight/A) x 100% 40 15 Pup weight (g) 30 10 20 5 10 0 0 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 h d 40 700 (Food intake/MFI) x 100% 35 600 30 Food intake (g) 25 500 20 400 15 300 10 5 200 0 100 0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21 Days in lactation Days in lactation 800 Cns Cs Sns Ss 600 400 200 0Figure 3 Average maternal body weight (3a), average litter weight (3b), average pup body weight(3c), average food intake (3d), average maternal body weight relative to A (3e), average litterweight relative to A (3f), average pup body weight relative to A (3g) and average food intakerelative to MFI (3h) for each standardisation level in each line from farrowing/birth to weaning.A = asymptotic mature virgin body weight (g); MFI = mature virgin food intake (g); C = controlline; S = selection line; ns = with non-standardised litters; s = with standardised litters.
    • 7. Food resource allocation patterns in lactating females 101heavier relative to A than dams with standardised litters from 19 to 21 days inlactation (P<0.05) (Figure 3e). From birth to weaning, litters of the S-line were heavier relative to A than littersof the C-line (P<0.001) and non-standardised litters were heavier relative to A thanstandardised litters (P<0.001) (Figure 3f). From 2 days in lactation to weaning, pups of standardised litters had a higherdegree of maturity than pups of non-standardised litters (P<0.001). From farrowing toweaning, pups of Cns-families had a higher degree of maturity than pups ofSns-families (P<0.001). From birth to one day in lactation, degree of maturity in pups ofCs-families was higher than degree of maturity in pups of Ss-families (P<0.01);afterwards degree of maturity was similar (Figure 3g). From farrowing to weaning, S-line families had a higher food intake relative to MFIthan C-line families; this was significant at 1, 4, 5, and 8 to 21 days in lactation (P<0.05).Families with non-standardised litters generally had a higher food intake relative toMFI than families with standardised litters but this was significant at 19 to 21 days inlactation only (P<0.001) (Figure 3h). Table 2 presents phenotypic correlations between several litter traits. Largerlitters had more stillborn pups and the pups were less mature at birth. Degree ofmaturity at birth was negatively correlated with number of stillborn pups andpre-weaning mortality rate. The day that the pups opened their eyes was later inanimals that were less mature at peak lactation (Table 2).Table 2 Phenotypic correlations between total number of pups born, and number of stillborn pupsand degree of maturity at birth, between degree of maturity at birth and number of stillbornpups and pre-weaning mortality rate, and between degree of maturity at peak lactation and theday that the pups open their eyes. Total number Number Pre-weaning Eyes open (d) pups born stillborn pups mortality rateNumber stillborn pups 0.36d ***Degree of maturity -0.56ad *** -0.23ad ** -0.35acd *** -0.30be ***a degree of maturity at birth; bdegree of maturity at peak lactation; cnon-standardised littersonly; dadjusted for line; eadjusted for line and standardisation; **P<0.01; ***P<0.001.3.2.2. Residual food intakeFigure 1 shows the average daily RFI for each standardisation level in each line fromfarrowing to weaning. R2 values of the multiple regressions according to equation (3)per day were in the range of 70% to 92%. Figure 1 shows that RFI from farrowing toweaning in Cns-families was 0. The reason for this is that the equation used to estimateRFI was based on all Cns-families and hence the average RFI of the Cns population was0. Figure 1 shows furthermore that there was not an explicit trend present for RFIduring lactation, as can be seen for RFI in non-reproductive females. Generally, RFI wasaround and above 0 for Cs-families, both above and below 0 for Sns-families, andaround and above 0 for Ss-families.
    • 102 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting Average RFI for each standardisation level in each line in the F-PL period and thePL-W period are presented in Figure 2. R2 values of the multiple regressions accordingto equation (3) per period were 93% for the F-PL period and 90% for the PL-W period. During the F-PL period, S-line families had lower RFI than C-line families (P<0.01).Families with standardised litters had higher RFI than families with non-standardisedlitters, but this was significant only in the S-line (P<0.001). During the PL-W period, Sns-families had lower RFI than C-line families (P<0.05) andSs-families (P<0.001). Ss-families had higher RFI than Cns-families (P<0.05) andSns-families (P<0.001). Families with standardised litters had higher RFI than familieswith non-standardised litters; this was significant only in the S-line (P<0.001).3.3. After weaning3.3.1. Body weight and food intakeTable 3 presents for each standardisation level in each line average body weight andfood intake from weaning to 25 days after weaning. S-line females were significantlyheavier than C-line females. Females with formerly non-standardised litters wereheavier than females with formerly standardised litters; this was significant in the S-line only. S-line females had a significantly higher food intake than C-line females. Food intakewas higher in females with formerly non-standardised litters than in females withformerly standardised litters, but this was significant in the S-line at 5 days afterweaning only.Table 3 Means and standard errors of body weight (BW) and food intake (FI) from weaning to25 days after weaning for each standardisation level in each line. days after weaning 5 10 15 20 25 a a a aBW (g) Cns 37.2 ± 0.497 38.1 ± 0.461 38.0 ± 0.433 37.0 ± 0.459 36.5 ± 0.402a Cs 36.3 ± 0.414a 37.7 ± 0.407a 37.7 ± 0.383a 36.7 ± 0.457a 36.6 ± 0.410a b b b b Sns 49.6 ± 0.564 50.7 ± 0.558 51.1 ± 0.567 50.5 ± 0.561 50.0 ± 0.513b Ss 48.2 ± 0.425c 49.0 ± 0.409c 49.3 ± 0.449c 48.3 ± 0.391c 48.4 ± 0.472cFI (g/d) Cns 6.47 ± 0.142a 6.03 ± 0.0762a 5.82 ± 0.104a 5.39 ± 0.0829a 5.34 ± 0.0748a a a a Cs 6.11 ± 0.108 5.97 ± 0.0950 5.71 ± 0.0659 5.22 ± 0.0737a 5.39 ± 0.0708a Sns 8.95 ± 0.155b 8.05 ± 0.118b 7.70 ± 0.0947b 7.10 ± 0.0983b 7.05 ± 0.106b Ss 8.31 ± 0.144c 7.81 ± 0.193b 7.43 ± 0.142b 6.99 ± 0.171b 6.93 ± 0.121bWithin a column, means without a common superscript letter differ (P<0.05). C = control line;S = selection line; ns = with non-standardised litters; s = with standardised litters. The results show that after weaning there was a decreasing trend in food intake,but not in body weight. At 25 days after weaning, body weight was on average 27%higher than A in Cns- and Ss-females and 28% higher than A in Cs- and Sns-females;these differences were not significant. At 25 days after weaning, food intake was on
    • 7. Food resource allocation patterns in lactating females 103average 14% higher than MFI in Cns, Sns and Ss-females and 15% higher than MFI inCs-females; these differences were not significant.3.3.2. Residual food intakeFigure 1 shows for each line the average RFI for each 5-day period from weaning to 25days after weaning (g/d). R2 values of the multiple regressions per day were in therange of 25% to 61%. Residual food intake was higher in S-line females than in C-linefemales (P<0.01). Figure 1 shows that RFI from weaning to 25 days after weaning inC-line females was 0. The reason for this is that the equation used to estimate RFI wasbased on all C-line females and hence the average RFI of the C-line population was 0. Average RFI per line for the ‘after weaning’ period is presented in Figure 2. The R2value of the multiple regression was 45%. Residual food intake during the ‘afterweaning’ period was significantly higher in S-line females than in C-line females(P<0.001).3.4. Correlation between residual food intake measurements in different periodsTable 4 presents phenotypic correlations between RFI in the growing period, the adultperiod, the F-PL period, the PL-W period and the ‘after weaning’ period. Residual foodintake in the growing period is highly correlated with RFI in the adult period. Residualfood intake from farrowing to peak lactation is highly correlated with RFI from peaklactation to weaning. Residual food intake in the non-reproductive period (i.e., thegrowing and the adult period) is not correlated with RFI during lactation (i.e., the F-PLand the PL-W period). Residual food intake after weaning is correlated both with RFI inthe non-reproductive period and RFI during lactation (Table 4).Table 4 Phenotypic correlations between residual food intake in the growing period, the adultperiod, the period from farrowing to peak lactation (F-PL), the period from peak lactation toweaning (PL-W) and the ‘after weaning’ period. growing perioda adult perioda F-PLb PL-Wbadult perioda 0.63 *** bF-PL 0.09 0.12PL-Wb 0.13 0.10 0.51 *** aafter weaning 0.38 *** 0.58 *** 0.32 *** 0.22 ** a b**P<0.01; ***P<0.001. adjusted for line; adjusted for line and standardisation.4. DiscussionObservations on body weight and food intake in non-reproductive C- and S-line femalesfrom 21 to 69 days of age in the present study support observations at earliergenerations that have been extensively discussed in Chapter 3: selection for high littersize has increased daily and mature body weight and food intake in non-reproductivemice as a correlated effect. Chapter 3 showed furthermore that growth and food
    • 104 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luitingintake curves against time differ significantly when standardised by estimates ofmature body weight. According to Taylor (1980) this indicates the presence of linespecific genetic factors, which suggests that selection for high litter size hasdisproportionally changed the allocation pattern of resources that are available to theprocesses of maintenance and growth (Chapter 3). Estimation of RFI is proposed as a tool to quantify resource allocation patterns(Chapter 6). Residual food intake is an estimate of the total amount of resourcesavailable to an animal for other processes than maintenance and production and istherefore suggested to be an estimate of the total amount of ‘buffer’ resources thatare left for, e.g., physical activity and the ability to cope with unexpected stresses(Luiting et al., 1997; Chapter 6). Residual food intake in non-reproductive males and females of the C- and S-line wasdiscussed earlier in Chapter 4; RFI in that study was estimated by regressing foodintake on metabolic body weight and body weight gain in the C-line population consistingof males and females. The results of that study showed that adult non-reproductiveindividuals of the S-line, in particular females of this line, have a significantly higherRFI than mice of the C-line (Chapter 4). Residual food intake in the present study wasestimated by regressing food intake on metabolic body weight and body weight gain inthe C-line population consisting of females only. The results of the present studysupport the observations presented in Chapter 4: adult non-reproductive S-line femaleshave significantly higher RFI than C-line females, suggesting that these females havemore ‘buffer’ resources left to respond to unexpected stresses and challenges. These buffer resources may be intended for the highly increased resourcedemanding processes of pregnancy and lactation. Increased energy for maintenancewith selection for heat loss in mice allowed for a greater litter size as a correlatedeffect in the study of Nielsen et al. (1997). Lactation has been noted as the period ofpeak energy demand and food intake for wild mammals (Oftedal, 1984). Weiner (1989)indicated that the maternal energy requirement at peak lactation can approximate themaximum for sustained work. In mice, Hammond and Diamond (1992) reported asustained metabolic rate of 7.2 times basal metabolic rate during lactation. Mammaryglands are the primary users of absorbed nutrients in lactating animals (Boyd andTouchette, 1997). Litter size is the major factor influencing milk production and istherefore the most important source of variation of the nutrient balance of the dam(Etienne et al., 1998). Since S-line dams have to support a litter that has practicallybeen doubled in size by selection, lactation may considerably change the resourceallocation patterns. For this reason, in the present study, food resource allocation, asquantified by RFI, was investigated in lactating females of the C- and S-line. To manipulate experimentally the energy burden of lactation, in each line, half of thefemales supported litters that were standardised at birth and half of the femalessupported all pups born. In the present study, litters were standardised to eight pupsper litter, when larger than eight pups. It should be noted that the difference in littersize between standardised and non-standardised litters was about four times larger inthe S-line than in the C-line. Energy intake increases greatly during lactation to acquire sufficient energy formaternal maintenance and milk production. In addition, body tissue mobilisation makes
    • 7. Food resource allocation patterns in lactating females 105an important contribution to the additional energy costs (Naismith et al., 1982; Pine etal., 1994). Rogowitz (1998) showed that about 90% of the metabolisable energyavailable to lactating cotton rats was obtained from food intake while the remainderwas derived from maternal stores. Food intake in the mice strain used in the study ofMillican et al. (1987) has been shown to rise to 4 times the virgin value by peaklactation. Hammond and Diamond (1992) observed an increase of 3.4 times virgin valuesat peak lactation and in the present study, dams of both lines reached an intake level ofaround 4 times their virgin mature food intake (MFI). Although Sns-dams supported atpeak lactation litters that were about 58% larger and, relative to A, 13% heavier thanCns-litters, food intake relative to MFI was only 10% higher than in Cns-dams. Aroundpeak lactation, the pups open their eyes, and the further increase in food intake can beattributed to both the dam and the offspring. Food intake relative to MFI decreasedafter weaning and the difference between the Sns- and Cns-dams disappeared. Food intake varied significantly with litter size in the study of Hammond andDiamond (1992): intake of dams with 14 pups (achieved by cross-fostering) was 25%higher than that of dams with five pups (natural size or achieved by culling). The meanrate of food intake was slightly but significantly higher in cotton rats with 6-puplitters (natural size) compared with dams with 3-pup litters (achieved by culling) in thestudy of Rogowitz and McClure (1995). In the present study, C-line dams withnon-standardised and standardised litters had similar food intakes up to peak lactation,while Sns dams eat significantly more than Ss dams during half of this period. This maybe due to the larger effect of standardisation on litter size in the S-line. Afterweaning, animals that weaned non-standardised and standardised litters had similarfood intakes. The amount of body tissue mobilisation in the study of Rogowitz (1998) wasestimated from maternal body weight loss during lactation; at peak lactation, extensivelosses of body fat and protein reserves may occur (Sainz et al., 1986). Also lactatingsows and dairy cows lose body weight during the lactation period, even under ad libitumfeeding conditions (Mullan et al., 1993; Tamminga et al., 1997). In the study ofHammond and Diamond (1992), body mass in lactating mice increased by 40% fromvirgin state through peak lactation and then did not change after weaning. Body weightsin lactating females of the present study seemed to follow the pattern of milkproduction: body weight increased up to peak lactation and subsequently decreased upto weaning. Indeed, a significant phenotypic correlation (adjusted for line andstandardisation; r = 0.28, P<0.001; results not presented) shows that the estimate ofday in lactation at peak lactation, estimated by fitting a quadratic function to data onmaternal body weight against days in lactation, is positively related to the day that thepups open their eyes. From that moment on, the pups start to eat solid food in additionto milk, which will progressively replace the contribution of milk to offspring growth. Inspite of the larger litter size and higher relative litter mass, body weights of Cns andSns dams increased to a similar level of more than 150% of their asymptotic matureestimates (A) at peak lactation; values decreased from peak lactation to weaning to138% in Cns dams and 145% in Sns dams. Within the first five days after weaning, bodyweights relative to A decreased further to about 127% in both lines and thereafter didnot change.
    • 106 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting Body mass and the rate of body weight loss did not vary with litter size in the studyof Rogowitz and McClure (1995). In the present study, body weight increasedsignificantly more in dams with non-standardised litters than in dams with standardisedlitters. After weaning, Sns dams were still heavier than Ss dams, but Cns and Cs damshad similar body weights. Since no weight loss occurred, no estimate can be given aboutuse of body stores during lactation. Forthcoming research will investigate body lipidmobilisation in lactating females of the C- and S-line. Since benefits to offspring have an associated maternal cost, trade-offs andconflicts may occur during lactation when a limit to food assimilation and sustainedmetabolic rate can be assumed to exist (Weiner, 1992; Rogowitz, 1996). If the damallocates too much of her resources to her offspring, she may lose weight excessively,increase her risk of mortality and compromise future reproductive potential; aninsufficient rate of energy export to young may decrease postnatal growth or causeoffspring mortality (Rogowitz, 1996). The observed negative relationship between littersize and pup weight and increase in pre-weaning mortality rates with larger litters(Hammond and Diamond, 1992; Rogowitz and McClure, 1995; the present study)indicates that a dam is to some extent able to protect her own stores at the expenseof the growing young. A higher investment of resources in lactation and the processesthat support this will result in a lower RFI as it was defined in this study. Preliminary results of the present study on the estimation of RFI in non-standardised mice of the C- and S-line are discussed in Chapter 8. Residual food intakewas then estimated by regressing food intake on maternal metabolic body weight,maternal body weight gain, litter size, pup metabolic body weight and pup body weightgain in Cns-families. However, the equation used in the present study, which includesthe two covariates ‘litter weight’ and ‘litter weight gain’, is a better model for which isa truly multiplicative system. Whereas daily RFI in the non-reproductive period followed a clear trend, the courseof RFI during lactation was rather capricious. A likely explanation may be that it tookconsiderably more time to weigh all dams and litters during lactation than to weight thefemales in non-reproductive state; the whole process took many hours. The dams andlitters were weighed in the same systematic order, but the older the pups became, themore time it took to weigh them all (they behaved like popcorn when the cage wasopened); the daily scheme was therefore quite irregular. Hammond and Diamond (1992)showed that food intake in lactating mice close to peak lactation rose in the afternoon,declined after midnight and was minimal at midday. Also, feeding times of the offspringmay differ. Since non-standardised litter size exceeds the number of teats (about 9 inthe C-line and 10 in the S-line), dams have been shown to solve this discrepancy bydividing the pups into two piles and nursing the piles alternately (Hammond andDiamond, 1992; the present study, data not presented). Therefore, during lactation,RFI estimated from accumulated data may be a better representation of the resourcesituation. From farrowing to peak lactation, RFI can be attributed to the dam only, while frompeak lactation to weaning, RFI can be attributed to both the dam and the pups. Residualfood intake from farrowing to peak lactation and from peak lactation to weaning wassignificantly lower in Sns-families than in Cns-families. This suggests that S-line dams
    • 7. Food resource allocation patterns in lactating females 107supporting litters of the size attained by artificial selection allocate more resources tothe processes that support milk production and have consequently fewer resources leftto respond to other demands. After weaning, RFI is significantly higher in S-linefemales than in C-line females, suggesting that, after weaning, the dams are able torestore the negative resource situation. From birth to peak lactation and from peak lactation to weaning, RFI was lower indams with non-standardised litters than in dams with standardised litters, though thiswas significant only in the S-line. Although litters of S-line dams were standardised torelatively smaller litters than litters of C-line dams, RFI was not significantly differentbetween Ss- and Cs-dams, as might have been expected; from peak lactation toweaning, RFI in Ss-families was higher than RFI in Cs-families, but this was notsignificant. After weaning, no differences were found between dams with formerlystandardised and non-standardised litters, nor when the equation used to estimate RFIfor the ‘after weaning’ period was based on the Cns-population (results not presented). Archer et al. (1998) found a moderate genetic correlation between post-weaning andmature RFI in non-reproductive mice and suggested that animals possess an ‘intrinsicefficiency’ that operates across different degrees of maturity and physiological states:the positive correlation results from basic physiological processes that are common toboth the growing animal and the mature animal, such as the absorption of nutrients.Lactation activates processes that are specific to the physiological state and (genetic)variation in these processes is unlikely to influence the efficiency of a non-reproductiveanimal (Archer et al., 1998). Results of the present study show that the phenotypiccorrelations between RFI in the non-reproductive period (i.e., the growing and the adultperiod) and RFI during lactation (i.e., the F-PL and the PL-W period) are very close tozero. This suggests that, during lactation, variation in milk production and theprocesses that support this dilute the importance of processes that are common tonon-reproductive and lactating animals as a source of variation in RFI. Indeed, thematernal body has to adapt greatly to the process of lactation. Apart from an increasein mammary size, lactating mice and rats experience an increase in liver, heart, lung andgut size to accommodate the large increase in food demands (Williamson, 1980; Pine etal., 1994; Speakman and McQueenie, 1996). In mice at peak lactation, Hammond andDiamond (1992) reported a 2.5-fold intestinal hypertrophy and Millican et al. (1987)observed a 2.8-fold increase in protein mass of the liver. Ferrell and Jenkins (1985)found that cattle with higher milk production have expressed higher maintenancerequirements independent of body mass; a large proportion of this variation wasexplained by critical organ mass, especially the liver. Milican et al. (1987) reported amore than 3 times increase in the rate of protein synthesis of the whole body at peaklactation. This was largely due to the synthesis by the mammary glands (38%), the liver(21%) and the gastrointestinal tract (24%). Phenotypic correlations between RFI afterweaning and RFI in the non-reproductive period are positive and highly significant,suggesting that common processes are again an important source of variation in RFI.Also phenotypic correlations between RFI after weaning and RFI during lactation arepositive and highly significant, indicating that processes that operate during lactationare still influencing the resource balance after weaning. This is plausible, since given
    • 108 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luitingthe aforementioned adaptations of the body to the process of lactation, it will taketime to return to non-reproductive state. Since tissues with high protein or high lipid levels have different maintenancerequirements, line and standardisation differences in body composition may explain partof the variation in RFI (Chapter 5). Protein turnover requires a high amount ofresources while body lipid is relatively metabolically inactive. Therefore, animals withrelatively high lipid content will have lower RFI than animals with relatively high proteincontent. Chapter 5 showed that body lipid and protein content is similar in C- and S-linefemales at 65 days of age and therefore does not likely explain the differences in RFIin adult non-reproductive state. Differences in body composition may influence RFIduring lactation when, e.g., the extent to which body reserves are mobilised isdifferent between the lines and standardisation levels, and furthermore largelyindependent of the traits which are included as covariates in the equation to estimateRFI. Furthermore, milk composition differs between different stages of lactation(Knight et al., 1986). Rogowitz and McClure (1995) observed that the milk produced bylactating cotton rats with large litters was dilute and had lower energy content per drymass than did the milk produced by animals supporting small litters. Since Sns-damshave to support a genetically highly increased litter size, these animals may mobilisemore body reserves and may produce more diluted milk than Cns-dams, and animals withnon-standardised litters may mobilise more body reserves and may produce milk of adifferent composition than animals with standardised litters. However, the degree ofbody tissue mobilisation may be positively correlated with litter size, which is includedas a covariate in the equation. Forthcoming research will investigate body composition inlactating females of the C- and S-line. Rogowitz (1996) observed a significant difference in the growth rate of offspring insmall and large litters of field-caught cotton rats. In that study, on average, individualpups in large litters (6 pups) grew at 71.2% the rate of pups in small litters (3 pups). Inthe present study, pup development in Sns-pups was about 25% lower than in Cns-pupsat all times. Interestingly, degree of maturity of Cs-pups and Ss-pups is similar fromtwo days in lactation on, which may indicate that the maximum relative growth rate issimilar in both lines and about 18% and 53% higher than Cns- and Sns-pups,respectively. This is furthermore supported by the aforementioned observation thatboth food intake and maternal body weight (as related to the level of milk production)up to peak lactation are lower in dams with standardised litters than in dams with non-standardised litters: a further increase is physically possible, but not used. The small litters in the study of Rogowitz (1996) were obtained by culling while largelitters were of natural size. Also in the present study, small litters in the C-line wereusually obtained by culling. This implies that also pups born in ‘natural sized’ litters (i.e.,non-standardised and non-selected) were under the influence of ‘maternal effects’, i.e.,limited by the maternal energy export in milk (Falconer and Mackay, 1996). Pigletgrowth rates during lactation remain, at best, half of which can be achieved underartificial rearing (Whittemore, 1980; Pluske and Dong, 1998). Though litter size in pigshas been increased by selection, this effect seems to result mainly from the relativelyhigh fat content and low protein content in sow milk. Piglets are born with a relativelylow body lipid content and under natural conditions priority is given to restore their
    • 7. Food resource allocation patterns in lactating females 109condition over improving their growth rate (Pluske and Dong, 1998). From birth to twoweeks of age, protein content of piglets increased from 12% to 15%, while fat contentincreased from 1.3% to 13% (Mullan et al., 1993). Whittemore (1980) suggested thatthe failure of the sow to supply adequately the nutritional needs of her offspring afterweaning may explain the curvature at the bottom end of the sigmoid shaped growthcurve and the subsequent long haul up the curve. Litter size may increase to the levelwhere dams can provide energy to offspring that allows for ‘sufficient’ offspringdevelopment. The present study shows that litter size in the S-line has been increasedbeyond this point: although S-line females with non-standardised litters allocate aparticularly high amount of resources towards the processes of lactation, this wasinsufficient to provide offspring with an adequate amount of resources, resulting inreduced pup development and increased pre-weaning mortality rates. To ensure that lactation proceeds successfully there are co-ordinated adaptations inthe metabolism (homeorhesis) that reallocate available nutrients towards the mammarygland away from tissues that are not essential to lactation (Bauman and Currie, 1980).It is generally observed that during lactation, self-maintenance of the dam takesprecedence over the maintenance of individual offspring, resulting in the dead ofoffspring under stressful conditions whereas the dam usually survives and subsequentlyreproduces (Rogowitz, 1996). However, single trait selection for high litter size mayresult in the situation where dams allocate ‘disproportionally’ many resources to thistrait selected for, leaving less resources to respond to other demands. In thatsituation, it is most likely that resources will be reallocated firstly from traits that arenot defined in the breeding goal, because they are given no importance (Chapter 2). Results of the present study suggest that dams selected for high litter size indeedallocated considerably more resources to maintenance of offspring than non-selecteddams. However, Sns-dams seemed to be able to restore the negative resource situationafter weaning. Because of increased food demands to support genetically increasedlitter sizes and reduced appetites and lower body fat reserves at parturition due togenetically increased leanness, the negative resource situation during lactation isgenerally more severe in commercial sows, than in the mice of the present study. Whena higher proportion of resources is allocated to lactation, less resources are left torespond adequately to other demands, putting the animal more at risk to behavioural,physiological and immunological problems (Bauman and Currie, 1980; Chapter 2). Indeed,commercial sows have frequent reproduction problems associated with excessivemobilisation of body reserves, such as prolonged weaning to oestrus intervals (TenNapel, 1996). Future research may investigate if lactating S-line females are indeedmore susceptible to stress and diseases and furthermore how the negative resourcesituation during lactation will affect lifetime reproduction potential. Mouse models,such as described in the present experiment, can be used to anticipate and preventundesirable side effects of selection in the long term.
    • 110 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. LuitingAcknowledgementsThis study is supported by a grant from the Norwegian Research Council, projectnumber 114258/111, ‘Consequences of selection for high litter size for the ability tocope in a stressful environment’. Kari Kjus is gratefully acknowledged for carrying outthe Norwegian mouse selection experiment and her help in providing and maintaining themice of this project. We thank January Weiner, Hans Ulrik Riisgård and ChristoferKnight for sending us their papers on request. This manuscript has been written at theInstituto Nacional de Investigación y Tecnología Agraria y Alimentaria (I.N.I.A.) inMadrid, Spain, which is thanked for providing the resources.
    • 7. Food resource allocation patterns in lactating females 111References• Archer, J.A. and Pitchford, W.S., 1996. Phenotypic variation in residual food intake of mice at different ages and its relationship with efficiency of growth, maintenance and body composition. Anim. Sci. 63:149-157.• Archer, J.A., Pitchford, W.S., Hughes, T.E., Parnell, P.F., 1998. Genetic and phenotypic relationships between food intake, growth, efficiency and body composition of mice post weaning and at maturity. Anim. Sci. 67:171-182.• Bauman, D.E. and Currie, W.B., 1980. Partitioning of nutrients during pregnancy and lactation: a review of mechanisms involving homeostasis and homeorhesis. J. Dairy Sci. 63:1514-1529.• Boyd, R.D. and Touchette, K.J., 1997. Current concepts in feeding prolific sows, Proc. 13th annual North Carolina swine nutrition conference.• Etienne, M., Dourmad, J.-Y., Noblet, J., 1998. The influence of some sow and piglet characteristics and of environmental conditions on milk production. In: M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (Eds.). The lactating sow, p. 285-299.• Falconer, D.S. and Mackay, T.F.C., 1996. Introduction to quantitative genetics, 4th edition, Longman Group Ltd, UK.• Ferrell, C.L. and Jenkins, T.G., 1985. Cow type and the nutritional environment: nutritional aspects. J. Anim. Sci. 61:725.• Hammond, K.A. and Diamond, J.D., 1992. An experimental test for a ceiling on sustained metabolic rate in lactating mice. Phys. Zool. 65:952-977.• Hanrahan, J. P. and Eisen, E.J., 1970. A lactation curve for mice. Lab. Anim. Care 20:101-104.• Knight, C.H., Maltz, E., Docherty, A.H., 1986. Milk yield and composition in mice: effects of litter size and lactation number. Comp. Biochem. Physiol. 84A:127-133.• Luiting, P., 1990. Genetic variation of energy partitioning in laying hens: causes of variation in residual feed consumption. World’s Poultry Sci. J. 46:133-152.• Jara-Almonte, M. and White, J.M., 1972. Milk production in laboratory mice. J. Dairy Sci. 55:1502-1505.• Luiting, P. and Urff, E.M., 1991. Optimization of a model to estimate residual feed consumption in the laying hen. Livest. Prod. Sci. 27:321-338.• Luiting, P., Vangen, O., Rauw, W.M., Knap, P.W., Beilharz, R.G., 1997. Physiological consequences of selection for growth. 48th Annual Meeting of the EAAP, Vienna, Austria.• Millican, P.E., Vernon, R.G., Pain, V.M., 1987. Protein metabolism in the mouse during pregnancy and lactation. Biochem. J. 248:251-257.• Mullan, B.P., Close, W.H., Cole, D.J.A., 1993. Predicting nutrient responses of the lactating sow. In: D.J.A. Cole, W. Haresign, P.C. Garnsworthy (Eds.). Recent developments in pig nutrition 2. Nottingham University Press, p. 323-346.• Naismith, D.J., Richardson, D.P., Pritchard, A.E., 1982. The utilization of protein and energy during lactation in the rat, with particular regard to the use of fat accumulated in pregnancy. Br. J. Nutr. 48:433-441.• Nielsen, M.K., Freking, B.A., Jones, L.D., Nelson, S.M., Vorderstrasse, T.L., Hussey, B.A., 1997. Divergent selection for heat loss in mice: II. Correlated responses in food intake, body mass, body composition, and number born through fifteen generations. J. Anim. Sci. 75:1469- 1476.• Oftedal, O., 1984. Milk composition, milk yield and energy output at peak lactation: a comparative review. In: M. Peaker, R.G. Vernon and C.H. Knight (Eds.). Physiological strategies in lactation. Symp. Zool. Soc. Lond. 51:33-85.
    • 112 W.M. Rauw, P.W. Knap, M.W.A. Verstegen and P. Luiting• Parks, J.R., 1982. A theory of feeding and growth of animals. Springer-Verlag, Berlin.• Pine, A.P., Jessop, N.S., Oldham, J.D., 1994. Maternal protein reserves and their influence on lactational performance in rats. Br. J. Nutr. 71:13-27.• Pluske, J.R. and Dong, G.Z., 1998. Factors influencing the utilisation of colostrum and milk. In : M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (Eds.). The lactating sow, p. 45-70.• Riisgård, HU, 1998. No foundation of a "3/4 power scaling law" for respiration in biology. Ecology Letters 1:71-73.• Rogowitz, G.L., 1996. Trade-offs in energy allocation during lactation. Amer. Zool. 36:197- 204.• Rogowitz, G.L., 1998. Limits to milk flow and energy allocation during lactation of the Hispid Cotton rat (Sigmodon hispidus). Phys. Zool. 71:312-320.• Rogowitz, G.L. and McClure, P.A., 1995. Energy export and offspring growth during lactation in Cotton Rats (Sigmodon hispidus). Funct. Ecol. 9:143-150.• Sainz, R.D., Calvert, C.C., Baldwin, R.L., 1986. Relationships among dietary protein, food intake and changes in body and tissue composition of lactating rats. J. Nutr. 116:1529-1539.• Statistical Analysis Systems Institute, 1985. SAS user’s guide: statistics, version 5, Statistical Analysis Systems Institute Inc., Cary, NC.• Speakman, J.R. and McQueenie, J., 1996. Limits to sustained metabolic rate: the link between food intake, basal metabolic rate, and morphology in reproducing mice, Mus Musculus. Phys. Zool. 69:746-769.• Tamminga, S., Luteijn, P.A., Meijer, R.G.M., 1997. Changes in composition and energy content of liveweight loss in dairy cows with time after parturition. Livest. Prod. Sci. 52:31-38.• Taylor, St C.S., 1980. Live-weight growth from embryo to adult in domesticated mammals. Anim. Prod. 31:223-235.• Taylor, St C.S. and Murray, J.I., 1987. Genetic aspects of mammalian growth and survival in relation to body size. Academic Press, London.• Ten Napel, J., 1996. Genetic aspects of intervals from weaning to estrus in swine. PhD thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Vangen, O., 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.• Weiner, J., 1989. Metabolic constraints to mammalian energy budgets. Acta Theriol. 34:3-35.• Weiner, J., 1992. Physiological limits to sustainable energy budgets in birds and mammals: ecological implications. Tree 7:384-388.• Whittemore, C.T., 1980. Elements of pig science. Longman Group UK.• Williamson, D.H., 1980. Integration of metabolism in tissues of the lactating rat. FEBS Letters 117:93-105.
    • 8 GENERAL DISCUSSION W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. VangenAbstract Reproductive failure contributes to more than 50% of total cullings in sowsafter first weaning. Prolonged rebreeding intervals are often associated with metabolicimbalance, disease or stress. Furthermore, a pre-weaning mortality of more than 20% isnot unusual and is mostly due to problems of development and adaptation. The presentstudy describes the consequences of selection for litter size for the allocation of feedresources, in relation to reproductive performance and pup development. Implicationsare illustrated by mice selection experiments. Average litter size in females from a lineselected for high litter size at birth (S-line) is about twice (20) the litter size offemales from a non-selected control line (C-line; 10). Lactating S-line females reallocatemore buffer resources towards lactation than lactating C-line females. Furthermore,they mobilise body stores for a longer period of time. This means that S-line femalesproduce more offspring but at a greater cost to their own metabolism. This processwas insufficient to supply offspring with adequate resources, resulting in reduced pupdevelopment and increased pre-weaning mortality rates. Results are discussed inrelation to pig production. Increased genetic gains will be compromised in the long-termif the short-term focus is on a single production trait only.Keywords: selection, litter size, resource allocation, fitness, reproductive performanceBased on Livestock Production Science 60, W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegenand O. Vangen, Selection for high production efficiency and its consequences for the allocationof feed resources - a concept and its implications illustrated by mice selection experiments, 329-341, ©1999 Elsevier Science B.V., with permission from Elsevier Science.
    • 114 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangen1. IntroductionSows are culled mainly for reasons involving reproductive failure, locomotion problemsor poor lactational performance; reproductive failure generally contributes to morethan 50% of total cullings after first weaning (Dourmad et al., 1994). A prolongedperiod of anoestrus after weaning is often found to be associated with metabolicimbalance, disease or stress (Ten Napel, 1996). Severe depletion of body reserves andstress may be mediated by factors such as nutritional deficiency, genotype, parity, milkproduction level, temperature, photoperiod, lactation length, litter size and socialenvironment (Prunier et al., 1996). Mammary glands are the primary users of absorbednutrients in lactating sows (Boyd and Touchette, 1997). Litter size is the major factorinfluencing milk production and is therefore the most important source of variation ofthe nutrient balance in lactating sows (Etienne et al., 1998). Weaning to conceptioninterval increased with the number of piglets nursed by 1 day/piglet in the study ofTholen et al. (1996). Another unfavourable relationship between improved production and reproductiveperformance is an increase in perinatal mortality rates. In pigs a pre-weaning mortalityof more than 20% is not unusual and occurs mostly during parturition and within thefirst 3 to 4 days of life; most of these losses are due to non-infectious causes andmany of these result from problems of development and adaptation (Svendsen, 1992). The present study aims at investigating the biological background of undesirableeffects of increased litter size on reproductive performance and development ofprogeny. Firstly, it is described how selection for high production efficiency maycompromise fitness. Thereafter, results are presented of a study into differences infood resource allocation patterns and reproductive performance in a long-termselection experiment for litter size in mice. The results are discussed in relation to pigproduction.2. Optimisation of biological design by natural selection2.1. The concept of symmorphosisWhat impressed Charles Darwin on his voyage around the world was his observationthat each species was so well adapted to its surroundings through adaptations ofexternal organs and behaviour. He developed the concept that competition betweenspecies and individuals for finite resources would eventually result in the gradualperfection of adaptations and the extinction of individuals or species unfit for theenvironment, i.e., "survival of the fittest" (Gottlieb, 1992). Does nature ‘optimise’ biological organisms? To answer this question, Taylor andWeibel (1981) proposed the hypothesis of optimal biological design or symmorphosis.Symmorphosis suggests an economic use of resources by all parts of a biologicalstructure (at any level, i.e., cells, organs, organisms), such that the parts arequalitatively and quantitatively coadjusted to their common role; never-used excess in
    • 8. General discussion 115biological capacity in any of the parts is costly in terms of maintenance, materials andspace and would therefore not be favoured by nature. It is mainly evolutionary biologists that have contested that evolution by naturalselection can lead to ‘optimal’ rather than merely ‘adequate, sufficient’ design,suggesting ‘survival of the fit’ more than ‘survival of the fittest’. Valid arguments arethat ‘organisms are not designed’ and ‘natural selection has no final goals and purposes’.Furthermore, ‘constraints and trade-offs are pervasive in biological systems’, ‘selectionis constrained to work with pre-existing materials (inherited alleles) and these may notbe the best possible materials for a particular function’, and ‘changes produced bymigration and genetic drift may not be adaptive’ (Gordon, 1998; Garland, 1998). Whentesting the hypothesis of symmorphosis at different levels of biological organisation,the more complex the system considered, the more results may be found that do notindicate perfect matching. A reason is that particular structures, e.g., entire organsystems, often serve multiple functions resulting in constraints and trade-offs. Ingeneral, the concept of symmorphosis holds mostly for internal characteristics of thebody whereas most problems arise in organs at the interface to the environment (suchas the gastrointestinal system). These organs must both adapt to the needs andconstraints of the internal environment and deal with the pressures and resistances ofthe variable and unpredictable external environment (Weibel, 1998). Consequently, adaptation to stresses of the environment, which is the primaryresult of natural selection, will never reach a state of perfect optimality, but may beconsidered as a ‘process of becoming’. It results in organisms ‘designed’ the best thatthey could be, given some explicit assumption of a design criterion and within specifiedconstraints (Garland, 1998).2.2. The resource allocation theoryAccording to the Resource Allocation Theory developed by Beilharz et al. (1993) theprevailing constraint to fitness is the availability of resources in the environment andthe efficiency of its use by the organism. The relationship is described as follows:fitness is a trait composed of several components, such as ‘number of parities’ and‘average litter size’, which multiply to give fitness. The resources consumed by theseand other processes (e.g., maintenance, (re)production, movement, reaction toenvironmental stresses) add to give the total amount of resources consumed(originating from, e.g., food intake, body tissue), since resources consumed by oneprocess are no longer available for other processes. As a consequence, in a resourcelimited environment, fitness will decrease if one of its components increases incombination with an increased allocation of resources to this trait. Natural selectionwill maximise fitness with optimal intermediate values for its components (Beilharz etal., 1993). If more resources become available by improvement of the environment,natural selection will raise fitness in a population whenever there is any additivevariation in any fitness component until the new environmental limit is reached (Beilharzand Nitter, 1998). The theory implies that different environments place different loads on genotypesof a population, and consequently have different optima for the allocation of resources
    • 116 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangento meet the different challenges provided by each environment. Secondly, differentgenotypes place different demands on environmental resources. In nature those animalswith genotypes producing phenotypes that use the available resources in the mostefficient way will produce more offspring than those that do not: animals with a highergenetic potential for fitness will produce more offspring than animals with lowergenetic potential. However, animals with a genetic potential that is too high will be lessfit because the environment is insufficient and unable to support the resource demand.As a result, different genotypes have been favoured and perform best in differentregions (e.g., Bos indicus cattle in the tropics and Bos taurus cattle in Europe), asituation termed ‘genotype-environment interaction’ (Beilharz and Nitter, 1998).According to the theory, it is the environment that allows or prevents evolutionarychanges to occur (Beilharz, 1998).2.3. Buffer capacitiesApart from genetic changes that may not keep pace with environmental changes, anindividual organism may have to deal with a variety of environmental conditions in itslifetime: when it is perfectly adapted to the summer, it may not survive the winter.Consequently, several non-extreme genotypic combinations will exist that producephenotypes that are as adequate, sufficient, and fit as required to live and reproduce.The genetic diversity that these individuals in a population represent may act as abuffering effect, allowing species to change genetically if the conditions alter (Lerner,1954). Physiological flexibility, or in other words ‘safety factors’, allow the individual tocope with environmental variability. Safety factors are defined as the ratio of acomponent’s capacity (strength, performance) to the normal expected load upon thatcomponent. Organisms often posses capacities somewhat in excess of what theynormally use, which makes it possible to withstand short-term stresses and adapt tolong-term changes in the environment (Hammond et al., 1994). Animals are able toincrease allocation of environmental food resources to respond to conditions such ascold stress and reproductive activity (Hammond et al., 1994). Fuel reserves stored inbody fat, protein and glycogen allow individuals to cope with variability in theavailability of food resources. Protein turnover, defined as the continuous breakdownand replacement of cellular proteins provide the flux that is necessary for metabolicregulation and adaptation (Hawkins, 1991). Homeostatic regulating mechanisms allowanimals to maintain constancy of several physiological conditions under moderateenvironmental changes (Withers, 1992). Load-bearing structures such as bones andtendons have more strength than is normally required, preventing these structuresfrom breaking too easily. Although buffer capacities are required to cope with (unpredictable) stresses,cost-benefit considerations may eventually decide whether and to which extent theyare maintained by natural selection. Flexibility is needed but costly.
    • 8. General discussion 1172.4. Consequences of artificial selection and genetic engineeringDomestication has removed several environmental constraints, such as variability inclimate and food quality and availability, and resource demanding processes such assearching for food and fleeing from predators. Both processes have increased theamount of available environmental resources considerably, which can thus be utilised byother resource demanding processes (Beilharz et al., 1993). Consequently, fitness isexpected to rise whenever there is any additive variation in any fitness component untilthe new environmental limit is reached (Beilharz and Nitter, 1998). However, with artificial selection in livestock species the optimal goal, which innature is highest fitness, has been redefined towards the ‘breeding goal’, which isdominated by high production. As a result, the resources that have become availablewith domestication are distributed mainly towards production traits and thoseprocesses that support this (Chapter 2). This process is suggested to have resulted inthe unprecedented increase in production levels of commercial livestock species duringthe past decades without strong evidence of constraints to genetic change (Beilharz etal., 1993). However, the numerous examples where artificial selection has resulted inanimals that are ‘unfit’, i.e., more at risk to behavioural, physiological and immunologicalproblems (Chapter 2) do suggest that we must expect livestock species to have becomeagain limited by their environment in many situations. When selection is forced beyond this limit the situation may arise in which also‘buffer resources’ are allocated towards production and insufficient buffer capacity isleft to respond adequately to unexpected stresses and challenges. This is likely sinceselection is mostly for high production efficiency, i.e., high production combined withlow food costs; food costs for other processes than production are economicallyunfavourable to the producer. An example is protein turnover: it is essential for lifebut it is also expensive. Breeding programmes have resulted in slower protein turnoverin the isolated tissues of rats and chickens selected for increased growth rate leadingto lower energy expenditure and higher growth efficiencies (Bates and Millward, 1981;Hayashi et al., 1985; Jones et al., 1986; Hawkins, 1991; Tomas et al., 1991). Otherenergetic trade-offs may concern immune response (Demas et al., 1997) and behaviour(Lachmansingh and Rollo, 1994). Preferential allocation of resources will occur when the demand of resourcesexceeds its availability. This situation of a trait leaving limited resources for theexpression of another trait is characterised by the occurrence of negative geneticcorrelations among traits (Luiting, 1999). When genetically ‘forced’ to produce highly,resources will be reallocated from other processes, leaving the animal lacking in abilityto respond to other demands. In this situation it is most likely that resources will bereallocated firstly from traits not defined in the breeding goal, because they are givenno importance (Chapter 2). Finally, when the manipulated trait is part of a multi-component system, and indeedmost systems are multifunctional, symmorphosis predicts that natural selection willbring the capacities of other components of the system into register with itsassociated capacity (Feder, 1998). However, if genetic changes are too radical orsought too rapidly the population may lack the time required to adapt to the changes
    • 118 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangenand homeostatic balance of the animal is at risk (Dunnington, 1990). This effect may bemore serious when introducing new genetic material with gene transfer or gene therapyinto a system that is in balance (e.g., Pursel et al., 1989; Massoud et al., 1996).3. Consequences of selection for high litter size for the allocation of food resources3.1. Material and methodsTo investigate consequences of selection for high litter size for food resourceallocation, several experiments have been conducted with mice allocated from twoselection lines of the Norwegian mouse selection experiment (e.g., Vangen, 1993): a lineselected for more than 100 generations, and since many generations plateaued, for highlitter size at birth (S-line) and an un-selected control line (C-line). Average number oflive-born pups in the S- and C-line is about 20 and 10, respectively. The mice are housedin cages of 30 x 12.5 x 12.5 cm3 filled with a bottom of saw dust and have free accessto pellet concentrate and water. The food contains 12.6 kJ ME per gram and 21% crudeprotein, as specified by the producer. The light is left on 24 hours a day. All mice,except for the offspring reported on in section 3.1.2, were raised from birth toweaning in litters standardised, when larger than eight pups, to eight pups per litter.3.1.1. Food resource allocation in non-reproductive femalesThe material and methods of this experiment are extensively described in Chapters 3and 4. Briefly, per line, 10 females in the 107th generation (replicate 1) and 16 femalesin the 110th generation (replicate 2) are randomly chosen at three weeks of age (i.e., atweaning) and housed individually. From 3 to 10 weeks of age, individual body weight (g)and food consumption (g) are measured three to five times a week. A Brody (1945) growth curve is fitted to individual data on body weight againstage, yielding, among others, individual estimates of asymptotic mature (virgin) bodyweight (A in kg). A linear function is fitted to relate individual data of cumulative foodintake to age, yielding individual estimates of mature daily food intake (MFI in kg/day).Brody curves and linear food intake functions are subsequently scaled according toTaylor (1980) by individual estimates of mature body weight A (kg): age and time curvevariables are divided by A0.27 and cumulated curve variables (such as body weight andcumulative food intake) are divided by A (Chapter 3). RFI is calculated, according to Luiting and Urff (1991), from multiple linearregression of food intake on metabolic body weight and body weight gain. RFI in thepresent experiment is defined as the difference between the food that is consumed byan animal and its consumption as predicted from requirements for growth andmaintenance per metabolic kg of the C-line population (the model is based on theaforementioned 10 C-line females, and 10 C-line males) (Chapter 4). The average RFI ofthis C-line population equals zero.
    • 8. General discussion 119 Body composition is measured in 30 females per line in the 116th generation, at 65days of age. The animals are starved 10 hours before killing but have free access towater. Animals are killed with CO2 and stored at -20°C. Before mincing all animalsindividually in a blender, the mice are thawed, and boiled for 10 minutes in a glas jar inwater. Minced samples are stored at -20°C until Kjeldahl N and lipid content areanalysed (unpublished results).3.1.2. Food resource allocation in reproductive females and its consequences for pup developmentPer line, in the 119th generation, 98 females are randomly chosen at three weeks of age(i.e., at weaning) and housed individually. From 3 to 10 weeks of age, individual bodyweight (g) and food consumption (g) are measured every three days. For each individual,A and MFI are estimated as reported on in section 3.1.1. At 10 weeks of age all females are mated and stay with the male for 2 weeks. Fromthe 87 C-line females and 96 S-line females that have become pregnant, the litters of42 dams of the C-line and 48 dams of the S-line are not standardised at birth. Fromfarrowing to weaning 2 dams of the C-line and 1 dam of the S-line died. Maternal bodyweight, litter weight, number of pups and food intake are measured daily fromfarrowing to weaning. Furthermore, the day that the pups open their eyes is recorded. RFI during lactation is calculated, according to Luiting and Urff (1991), from multipleregression of food intake on maternal metabolic body weight, maternal body weightgain, pup metabolic body weight, pup body weight gain and litter size. RFI in thepresent experiment is defined as the difference between the food that is consumed byan animal and its consumption as predicted from requirements for growth andmaintenance of dam and offspring of the C-line (see Chapter 4). The average RFI ofthe C-line dams equals zero. Degree of maturity is calculated as pup bodyweight / maternal A. Body composition is measured in 28 females per line in the 116th generation, at twoweeks in lactation and at weaning. Procedures are as reported on in section 3.1.1.3.2. Results and discussion3.2.1. Food resource allocation in non-reproductive femalesFigure 1a presents Brody (1945) growth curves and linear food intake functions of non-reproductive females of both lines (Chapter 3). Selection for high litter size hasincreased both body weight and food intake variables considerably. Table 1 presentsleast squares means estimates (adjusted for effect of replicate) of mature bodyweight (A) and mature daily food intake (MFI). Both estimates are about 35% higher inS-line mice than in C-line mice (Chapter 3). Since selection for high body weight usuallyleads to larger litters (Taylor and Murray, 1987), these results suggest that selectionfor high litter size has resulted in proportionally larger animals which consequently eatmore and produce larger litters.
    • 120 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangen a 45 350 Cumulative feed intake (g) 40 300 Body weight (g) 35 250 30 200 25 150 20 100 15 50 10 0 20 30 40 50 60 70 20 30 40 50 60 70 Age (d) Age (d) b Cumulative feed intake / A 1.0 10 0.9 8 Body weight / A 0.8 0.7 6 0.6 4 0.5 2 0.4 0.3 0 80 130 180 230 80 130 180 230 0.27 0.27 Age / A Age / A C SFigure 1 Average Brody growth curves and linear food intake functions (a) and average scaledBrody growth curves and linear food intake functions (b) per line (Chapter 3). All curves arebased on least squares means of curve parameters per line adjusted for effect of replicate.N = 26; C = control line; S = selection line; A = mature body weight (kg). A simple method to test this hypothesis is the genetic size-scaling theorydeveloped by Taylor (1980). According to the theory, if the hypothesis holds, thecorrelated response in body weight and food intake must disappear after correctionfor mature body weight (A). Figure 1b shows the effect of scaling growth curves andlinear food intake functions according to Taylor (1980) (Chapter 3): althoughdifferences between the lines in body weight and food intake variables decreasegreatly after scaling, substantial differences remain. According to Taylor (1980), theremaining differences are a consequence of line specific genetic factors (SGFs), whichindicates that selection for high litter size has done more than merely changing themature size. These SGFs may result from a situation where a further increase ingenetic size is limited by a limited uptake of resources from the environment. Selectionof litter size beyond this point may considerably change the allocation of metabolicresources towards maintenance and production (Luiting, 1999). Residual food intake (RFI) can be used as a tool to quantify these SGFs. An animalwith a negative RFI has consumed less food than predicted by the model, and hence ismore food efficient than the average of the C-line population on which the model is
    • 8. General discussion 121formed. Variation in RFI can be caused by variation in partial efficiencies formaintenance and growth, and by variation in metabolic (i.e., food demanding) processesnot included in the model, such as physical activity, responses to pathogens and stress(Luiting and Urff, 1991). Hence, this implies that RFI can be used to deal with resourceallocation issues and is a measure of the available buffer resources (Luiting, 1999). Table 1 presents least squares means (adjusted for effect of replicate) of RFIduring growth (from 3 to 6 weeks of age) and at maturity (from 6 to 10 weeks of age).In growing mice, RFI was not significantly different between C-line and S-line females.In adult mice, S-line females had higher RFI (are less food efficient) than C-linefemales (Chapter 4). Because growth is virtually absent, differences in RFI in adultanimals are mainly explained by differences in maintenance requirements (Luiting andUrff, 1991).Table 1 Least squares means per line (adjusted for effect of replicate), for mature virgin bodyweight (A), mature daily food intake (MFI), residual food intake (RFI) during growth and atmaturity, and means and standard errors (between brackets) of absolute protein mass andabsolute lipid mass. C SA (g) 31.21 42.95 ***MFI (g) 4.86 6.63 ***RFI during growth (g/d) 0.09 0.21RFI at maturity (g/d) 0.08 0.81 ***Absolute protein mass (g) 5.02 (0.05) 7.05 (0.11) ***Absolute lipid mass (g) 3.35 (0.23) 4.75 (0.25) ***C = control line; S = selection line; *P<0.05; **P<0.01; ***P<0.001. The results show that the line differences in specific genetic factors in non-reproductive female mice originate from differences in maintenance requirements:selection for high litter size has resulted in females with higher maintenancerequirements. The higher RFI in S-line mice shows that these animals have moreresources available for processes such as response to environmental stress than C-lineanimals (Chapter 4).3.2.2. Food resource allocation in reproductive females and its consequences for pup developmentIt is interesting that particularly adult females of the S-line increase their RFI, sinceit is these animals that can express the trait their genotype has been selected for. Ithas been suggested that the increase in RFI in this stage of life may be an anticipationto the metabolically highly stressful periods pregnancy and lactation (Chapter 4).However, according to the Resource Allocation Theory by Beilharz et al. (1993), greatlyincreased litter sizes, i.e., greatly increased resource demands, by means of artificialselection may drastically change the resource allocation pattern. In a resource limited
    • 122 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangenenvironment this may result in the situation where buffer resources and resources forprocesses other than reproduction are reallocated towards pregnancy and lactation. Table 2 presents average body weight and average litter weight divided by averagematernal virgin body weight (A) and average daily food intake divided by mature dailyfood intake (MFI), at farrowing, peak lactation (14 days) and at weaning (21 days).Furthermore, the table presents average absolute body protein and lipid mass and RFIfrom farrowing to peak lactation and from peak lactation to weaning. Average bodyweight/A at farrowing is very similar between C- and S-line dams; S-line dams,however, produce a significantly heavier litter proportional to maternal A. In lactatingfemale mice, body mass increases from farrowing to peak lactation to provide milkenergy for offspring growth. This increase is slightly more in S-line dams than in C-linedams, though non-significantly. After about 14 days, the pups begin to eat solid foodand as a consequence maternal milk production and maternal body weight decrease. Atweaning, average body weight/A is significantly lower in C-line dams compared withS-line dams. Litter weight/A increases up to weaning and remains significantly higher inthe S-line than in the C-line. Absolute body protein mass in lactating females at peaklactation is 39% higher in C-line dams and 42% higher in S-line dams as compared withvirgin females of each line (for virgin body composition see Table 1). The difference inabsolute body protein mass between lactating and virgin females decreases from peaklactation to weaning to 33% in C-line females and 38% in S-line females. These resultsshow that milk production in S-line dams is relatively higher and more persistent thanin C-line dams. Energy intake from farrowing to peak lactation increases greatly to acquiresufficient energy for maternal maintenance and milk production. In addition, toaccommodate this large increase in food demands, lactating mice experience an increasein liver, heart, lung and gut size (Speakman and McQueenie, 1996). Daily foodintake/MFI in C- and S-line dams increases greatly from farrowing to peak lactation;S-line dams eat proportionally more than C-line dams. Food intake from peak lactationto weaning can be attributed to both the dam and the pups. Because of the largerS-line litters, food intake/MFI increases particularly much in this line. Even thoughS-line dams increase their food intake more than C-line dams, RFI from farrowing topeak lactation is significantly lower in S-line dams. RFI from peak lactation to weaning,which can be attributed to both the dam and her offspring, was still significantly lowerin S-line than in C-line families, though the difference was greatly reduced. In additionto food intake, body fat is mobilised to provide for necessary energy: in both linesabsolute lipid mass at peak lactation is 57% lower compared with virgin females. Whilethe difference in absolute body lipid mass between lactating and virgin C- line femalesdecreased to 43% from peak lactation to weaning, the difference between S-linelactating and virgin females remained unchanged. A distinct exception of the principle of Taylor’s (1980) genetic size-scaling rules isgestation length, of which the difference between large and small sized breeds within aspecies is negligible. Taylor suggests that this results from the inherent feature ofbreeds within a species to be able to reproduce. Whereas gestation length, maturity atbirth, and litter size are independent of A, litter weight is proportional to A0.83, anobservation that is defined as the ‘constant maternal capacity’. Within a breed, an
    • Table 2 Means and standard errors (between brackets) of ‘body weight/A’, ‘litter weight/A, ‘food intake/MFI’, residual feed intake (RFI) fromfarrowing to peak lactation and from peak lactation to weaning, absolute protein mass, absolute lipid mass and degree of maturity (‘pup weight / A’). Farrowing/Birth Peak lactation Weaning C S C S C SBody weight / A (g) 1.32 (0.02) 1.30 (0.01) 1.52 (0.02) 1.56 (0.02) 1.38 (0.02) 1.45 (0.02) *Litter weight / A (g) 0.58 (0.02) 0.85 (0.02) *** 2.24 (0.08) 2.53 (0.05) ** 3.17 (0.12) 3.95 (0.09) ***Feed intake / MFI (g/d) 1.72 (0.08) 2.05 (0.07) ** 3.88 (0.16) 4.28 (0.09) * 4.66 (0.18) 6.12 (0.12) ***RFI (g/d) 0.00 (0.11) -1.69 (0.24) *** 0.00 (0.11) -0.75 (0.29) *Absolute protein mass (g) 6.96 (0.11) 10.02 (0.12) *** 6.70 (0.08) 9.70 (0.13) ***Absolute lipid mass (g) 1.43 (0.08) 2.05 (0.12) *** 1.90 (0.13) 1.96 (0.11)Degree of maturity 0.057 (0.001) 0.042 (0.001) *** 0.259 (0.009) 0.189 (0.006) *** 0.365 (0.011) 0.297 (0.010) ***C = control line; S = selection line; A = mature body weight; MFI = mature daily food intake; *P<0.05; **P<0.01; ***P<0.001. 8. General discussion 123
    • 124 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangenincrease in litter size must be compensated for by a decrease in individual birth weightfor the maternal capacity to remain at a constant level. Similarly, young of largerbreeds have less time to develop to a given degree of maturity when gestation length isfixed and are consequently born at a less mature age. Maternal capacity in large breedsis usually achieved by increased litter size. Degree of maturity at birth is related tofitness, since less mature pups have lower chances to survive (Taylor and Murray,1987). In our mice data, dams with largest litters have offspring that are less mature atbirth (r = -0.56; P<0.001) and have a higher number of stillborn pups (r = 0.36; P<0.001).Table 2 presents degree of maturity of pups at birth, peak lactation and at weaning.Degree of maturity at birth is lower in S-line pups than in C-line pups. The observednumber of stillborn pups is higher in S-line (0.67) than in C-line litters (0.29; P<0.05). Itis, however, not a very accurate estimate since dams have been seen to eat all stillbornpups within half an hour after farrowing. Pre-weaning mortality in S-line litters is considerably higher (36%) than in C-linelitters (18%; P<0.001). Although, as aforementioned, litter weight/A up to weaning ishigher in S-line mice, pup development in S-line litters is about 25% lower than in C-linepups at all times. Pups in larger litters open their eyes at a later age than pups fromsmaller litters (r = 0.43; P<0.001); C-line pups open their eyes on average 1 day (at 13days in lactation) before S-line pups do.4. Conclusions and synthesisIt is desirable that breeding on reproductive traits results in offspring fit at weaning,and is not compromised by prolonged weaning to oestrus intervals, reduced longevity,decrease in birth weights or increased mortality rates. However, these trade-offsseem to prevail when reproduction levels are increased (Dourmad et al., 1994). The first law of thermodynamics, which recognises conservation of energy, preventsa female from producing and sustaining larger litters than she can energeticallysupport. Increasing litter size beyond the point which can be supported by intake offood must then result in reallocation of maternal requirements to offspring, or indiminished offspring development. The present study indicates that female miceselected for high litter size at birth do both. The lower RFI in S-line dams comparedwith C-line dams during lactation shows that these dams are more food efficient.Furthermore, whereas control dams regain body condition after peak lactation, bodycondition from peak lactation to weaning in selected females remains unchanged. Thisprocess, however, is still insufficient to supply offspring with adequate resources,resulting in reduced pup development and increased pre-weaning mortality rates.Reallocation towards reproductive performance of buffer resources that are otherwiseavailable for processes such as physical activity, responses to pathogens and stress, willput the animal more at risk to behavioural, physiological and immunological problems(Chapter 2) and may compromise future reproductive potential (Rogowitz, 1998). Milk production in sows has a higher priority than maintaining the sows body tissuesuntil the loss of lean or fat tissue becomes too great (Whittemore, 1994; Noblet et al.,1998). In contrast with the mice in the present study, whose body weights increase up
    • 8. General discussion 125to peak lactation with increasing milk production, even in the best managementconditions, lactating sows lose body weight during the lactation period, even under adlibitum feeding conditions (Lodge et al., 1961; Noblet et al., 1998). It is estimated thatthe mean lactation energy requirements of modern lactating sows range between 6 to 8kg of conventional feeds per day; food intakes that are hardly achieved under practicalconditions (Dourmad et al., 1994; Noblet et al., 1998). Due to highly increased(re)production levels of sows over the last decades, the negative resource situation incommercial sows is generally more severe than modelled by mice in the present study.The demand of food resources has increased as a result of increased levels of milkproduction (Mackenzie and Revell, 1998) and higher maintenance requirements becauseof an increase in mature size of commercial sows (Noblet et al., 1998). At the sametime, sows in commercial production systems that originate from genetically improvedstrains of lean pigs, may have decreased food intake capacity, reduced appetite andlower body fat reserves at parturition, i.e., reduced amounts of available resources(Ten Napel, 1996; Whittemore, 1996). The sow will mobilise body reserves in order tomaintain milk nutrient output, instead of conserving these reserves to ensure a directstart to the next reproductive cycle (Mullan and Williams, 1989). Hence, they exhibitmore frequent reproduction problems associated with excessive mobilisation of bodyreserves, such as prolonged weaning to oestrus intervals (Dourmad et al., 1994; Nobletet al., 1998). The negative resource situation is generally more severe in primiparous sows.Selection for increased growth rate, efficiency and leanness in pigs has resulted in anincrease of around 30% in mature size over 20 years (Whittemore, 1994). Gilts are nowbred at a lower proportion of their mature size and are challenged simultaneously withthe drive to grow, support pregnancy, sustain lactation and re-breed after weaning,resulting in lengthened re-breeding intervals and a reduced number of piglets in thesecond litter (Whittemore, 1994, 1996). Before the mouse pups open their eyes they are entirely dependent on maternalenergy from milk, and the effects are therefore known as ‘maternal effects’ (Falconerand Mackay, 1996). Pup birth weight and pup development ultimately reflect the dams’ability to acquire sufficient energy for pregnancy and lactation to maintain the litter.The ‘number of pups fit at weaning’ produced determines reproductive success, and isthe ultimate goal of improved litter size. The negative relationship between litter size and birth weight in our mouse data isalso found in pigs. With increased litter size, piglet birth weight decreased by 0.03 kg(Kerr and Cameron, 1995). In the same study, piglet pre-weaning mortality decreasedrapidly with increasing piglet birth weight. The maternal effect on pup development inour mouse data is easily shown by means of litter standardisation: degree of maturityof C-line and S-line pups in litters standardised to 8 pups per litter is similar from 2days in lactation on, and about 18% and 53% higher than C- and S-line pups in non-standardised litters, respectively (data not presented). Although daily milk production in sows has increased over the past 2-3 decades byapproximately 1-2% per year (Mackenzie and Revell, 1998), piglet growth rates duringlactation remain, at best, half of what can be achieved under artificial rearing (Pluskeand Dong, 1998). Milk intake per piglet nursed decreases in a linear manner when litter
    • 126 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangensize increases (Etienne et al., 1998). Several studies have investigated whetherlactational performance is limited by a central constraint, i.e., a general limit to theassimilation of energy resources, or peripherally at the site where the energy is used.Hammond et al. (1994) and Rogowitz (1998) showed that individual food intake inlactating female mice and rats, apparently at the limit of their lactational performance,increased when transferred to a cold environment: even though the rate of food uptakeseemed to approach a maximum during lactation, an apparent buffer capacity was leftwhich under normal conditions was not used for improved pup development. It wastherefore suggested that these results support the hypothesis that lactational limitsare controlled peripherally at the site of the mammary gland. However, the significantincrease in food intake during lactation is associated with increased heat increment offeeding. In ad libitum fed lactating females, such as in the present study, a limit toheat loss may thus set a central limit to the assimilation of food resources, i.e., thetotal amount of food resources available to a dam. The zone of thermal comfort for the lactating sow is between 12 and 22°C (Black etal., 1993). Higher temperatures will result in decreased milk production andconsequently declined piglet growth, resulting from decreased food intake to reduceheat production and changes in the direction of blood flow to facilitate heat loss (Blacket al., 1993; Makkink and Schrama, 1998). Hence, for the most efficient production,sows should be kept within their zone of thermal comfort (Williams, 1998). However,the zone of thermal comfort for piglets is often at least 10°C higher than that for thesow, resulting often under modern farrowing conditions in cold stress for the pigletswhile the sow is suffering heat stress (Williams, 1998). From the moment that the mouse pups open their eyes, they begin to eat solid food,although the litters are nursed up to weaning, and as a consequence compensatorygrowth to overcome early growth depression is possible. Monteiro and Falconer (1966)recorded that the variance in pup weight of mice caused by maternal effects increasedfrom birth to 4 weeks of age (i.e., one week after weaning) and then decreased byabout 60% between 4 and 8 weeks. Consequently, despite the compensatory growth,sexual maturity was reached at a later age in pups from larger litters. The fact thatfemales reared in larger litters have smaller litters at first parity (Falconer andMackay, 1986), may thus result from the fact that these females, when mated at afixed age, are less mature. It follows that food protein and energy amounts during pregnancy and lactation mustbe accurately adjusted to nutritional requirements in order to prevent severe loss ofbody condition during lactation and subsequent prolonged weaning to oestrus intervals(Mullan and Williams, 1989; Yang et al., 1989; Whittemore, 1996). It is well establishedthat the more a sow eats in pregnancy, the less she eats in lactation (Williams, 1998)and as a consequence, in contrast to the feeding policy applied to the mice in thepresent study, food intake in sows is generally restricted during pregnancy.Whittemore (1998) concludes, however, that for the modern genotype the positivebenefits of good body condition at farrowing to buffer nutritional stress duringlactation far outweigh the small negative effects that may exist consequent upon theinverse relationship between pregnancy and lactation food intakes.
    • 8. General discussion 127 When food requirements are met, a further increase in litter size may beaccomplished by increased maternal capacity. Selection for litter size beyond the pointwhich can be supported by energy intake and maternal capacity is possible only to theextent where weight of individual offspring is just sufficient to survive. Supplementarymilk feeding may offer a practical way to support piglet growth when the sow’s milkproduction is not sufficient (Pluske and Dong, 1998). From the discussion it is apparent that selection of commercial pigs for productiontraits, such as reduced backfat and increased food efficiency, is decreasing the sowsreproductive performance. In this context it will be interesting to explore thepossibilities for producing specialised nucleus populations in the future where highprolific lines with low production qualities, such as the Meishan breed, are utilised asembryo recipient or foster population for piglets with high genetic merit for leanproduction. Luxford and Beilharz (1990) showed that when selection was only for litter size atbirth, capacity for lactation did not improve. However, when selection was for littersize at weaning, both number born and number weaned rose, since an increase of littersize weaned must have been supported by sufficient lactational capacity. In theNorwegian mouse selection experiment, selection is for number born. Furthermore, thestandardisation back to eight pups per litter removes the selection pressure onlactation capacity. The increase in lactation performance in S-line females in thepresent study, which nurse non-standardised litters, may be a phenotypic response tothe litters’ greater energy demand which the dams produce at a great cost to their ownmetabolism. Similarly, in most species, increase in high litter size with selection isprimarily due to increased ovulation rates (Pérez Enciso and Bidanel, 1997). However,selection for ovulation rate in pigs and mice increases ovulation rate, but is accompaniedby an increment in prenatal mortality and hence does not increase litter size to asimilar extent (Blasco et al., 1993). Each foetus requires a certain minimum space orlength of uterus to survive and develop (e.g., Chen and Dziuk, 1993). Hence, whereaslitter size from conception to birth must have been supported by sufficient maternalcapacity and energy allocation, maternal support has not been given weight whenselection is for ovulation rate only, resulting in the observed absence of response inlitter size at birth. Selection for increased litter size should therefore aim atimproving piglet growth rates, and litter size and pup weight at weaning. It can be debated whether experimental results obtained in mice can beextrapolated to pigs. Mice have the great advantage of having short generationintervals and are small and easy to handle, such that traits can be measured on largenumbers of animals which are raised under similar environmental conditions at relativelylow costs. Furthermore, several aspects of mouse studies have been verified inlivestock; e.g., maternal variation influencing body weight in the pig (Eisen, 1974).However, extrapolation to pigs needs care and results obtained in mice may need to beverified in pigs. Preliminary results obtained in mice can make final experiments in pigsmore efficient and less expensive when experimental design can be supported by apriori information, such as expected within- and between-treatment (co-)variances. Forcertain problems, such as long-term effects of selection, the only mammalian dataavailable may come from the mouse (Eisen, 1974). Long-term mouse selection
    • 128 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangenexperiments can be used to anticipate and prevent undesirable side effects ofselection. This chapter has presented a theory on the relation between food resourceallocation and undesirable side effects of selection for litter size. To generateimproved production without meeting concomitant requirements can merely result intrade-offs. Because of the interrelationships of physiological processes in function andshare of resources, increased genetic gains will be compromised in the long-term if theshort-term focus is on a single production trait only. Direction and response toselection cannot be accurately predicted without knowing the underlying physiologicalprocesses on which genetic selection acts. The ultimate knowledge is stored in thegenes and a lot of new information will be available in the future. Only when biologicalbackgrounds are understood may negative side effects be properly anticipated andprevented.AcknowledgementsThis study is supported by a grant from the Norwegian Research Council, projectnumber 114258/111, ‘Consequences of selection for high litter size for the ability tocope in a stressful environment’. Kari Kjus is gratefully acknowledged for carrying outthe Norwegian mouse selection experiment and her help in providing and maintaining themice of this project. Pieter Knap and the anonymous reviewers are greatly thanked fortheir useful comments on the manuscript. One anonymous reviewer is thanked for hissuggestion of exploring the possibilities for producing specialised nucleus populations inpigs.
    • 8. General discussion 129References• Bates, P.C. and Millward, D.J., 1981. Characteristics of skeletal muscle growth and protein turnover in a fast-growing rat strain. Br. J. Nutr. 46:7-13.• Beilharz, R.G., 1998. Environmental limit to genetic change. An alternative theorem of natural selection. J. Anim. Breed. Genet. 115:433-437.• Beilharz, R.G., Luxford, B.G., Wilkinson, J.L., 1993. Quantitative genetics and evolution: Is our understanding of genetics sufficient to explain evolution? J. Anim. Breed. Genet. 110:161-170.• Beilharz, R.G. and Nitter, G., 1998. The missing E: The role of the environment in evolution and animal breeding. J. Anim. Breed. Genet. 115:439-453.• Black, J.L., Mullan, B.P., Lorschy, M.L., Giles, L.R., 1993. Lactation in the sow during heat stress. Livest. Prod. Sci. 35:153-170.• Blasco, A., Bidanel, J.P., Bolet, G., Haley, C.S., Santacreu, M.A., 1993. The genetics of prenatal survival of pigs and rabbits: a review. Livest. Prod. Sci. 37:1-21.• Boyd, R.D. and Touchette, K.J., 1997. Current concepts in feeding prolific sows. Proc. 13th annual North Carolina swine nutrition conference.• Brody, S., 1945. Bioenergetics and growth. Reinhold, New York.• Chen, Z. -Y. and Dziuk, P.J., 1993. Influence of initial length of uterus per embryo and gestation stage on prenatal survival, development, and sex ratio in the pig. J. Anim. Sci. 71:1895-1901.• Demas, G.E., Chefer, V., Talan, M.I., Nelson, R.J., 1997. Metabolic costs of mounting an antigen-stimulated immune response in adult and aged C57BL/6J mice. Am. J. Phys. 42:R1631- R1637.• Dourmad, J.Y., Etienne, M., Prunier, A., Noblet, J., 1994. The effect of energy and protein intake of sows on their longevity: a review. Livest. Prod. Sci. 40:87-97.• Dunnington, E.A., 1990. Selection and homeostasis. Proc. 4th World Congr. Gen. Appl. Livest. Prod., Edinburgh, UK (XVI), 5-12.• Etienne, M., Dourmad, J.-Y, Noblet, J., 1998. The influence of some sow and piglet characteristics and of environmental conditions on milk production. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 285-299.• Falconer, D.S. and Mackay, T.F.C., 1996. Introduction to quantitative genetics. 4th edition., Longman Group Ltd, UK.• Feder, E.F., 1998. Testing the evolutionary origin and maintenance of symmorphosis. In: Principles of animal design. The optimization and symmorphosis debate. Weibel, E.R., Taylor, C.R., Bolis, L (ed.). Cambridge University Press, p. 48-55.• Garland, T., 1998. Conceptual and methodological issues in testing the predictions of symmorphosis. In: Principles of animal design. The optimization and symmorphosis debate. Weibel, E.R., Taylor, C.R., Bolis, L (ed.). Cambridge University Press, p. 40-47.• Gordon, M.S., 1998. Evolution of optimal systems. In: Principles of animal design. The optimization and symmorphosis debate. Weibel, E.R., Taylor, C.R., Bolis, L (ed.). Cambridge University Press, p. 37-39.• Gottlieb, G., 1992. Charles Darwin on the evolution of species and the role of embryological development. In: Individual development and evolution. The genesis of novel behavior. Oxford University Press, p. 19-26.• Hammond, K.A., Konarzewski, M., Torres, R.M., Diamond, J., 1994. Metabolic ceilings under a combination of peak energy demands. Phys. Zool. 67:1479-1506.• Hawkins, A.J.S., 1991. Protein turnover: a functional appraisal. Funct. Ecol. 5:222-233.
    • 130 W.M. Rauw, P. Luiting, R.G. Beilharz, M.W.A. Verstegen and O. Vangen• Hayashi, K., Tomita, Y., Maeda, Y., Shinagawa, Y., Inoue, K., Hashizume, T., 1985. The rate of degradation of myofibrillar proteins of skeletal muscle in broiler and layer chickens estimated by N tau-methylhistidine in excreta. Br. J. Nutr. 54:157-163.• Jones, S.J., Aberle, E.D., Judge, M.D., 1986. Skeletal muscle protein turnover in broiler and layer chicks. J. Anim. Sci. 62:1576-1583.• Kerr, J.C. and Cameron, N.D., 1995. Reproductive performance of pigs selected for components of effient lean growth. Anim. Sci. 60:281-290.• Lachmansingh, E. and Rollo, C.D., 1994. Evidence for a trade-off between growth and behavioral activity in giant supermice genetically-engineered with extra growth-hormone genes. Can. J. Zool. 72:2158-2168.• Lerner, I.M., 1954. Genetic homeostasis. John Wiley & Sons, Inc., New York.• Lodge, G.A., McDonald, I., MacPherson, R.M., 1961. Weight changes in sows during pregnancy and lactation. Anim. Prod. 3:269-253.• Luiting, P., 1998. The role of genetic variation in feed intake and its physiological aspects: results from selection experiments. In: Regulation of Feed Intake. Proceedings of the 5th Zodiac Symposium, Wageningen, The Netherlands (ed. D. van der Heide, E.A. Huisman, E. Kanis, J.W.M. Osse and M.W.A. Verstegen), CAB International, Wallingford. p. 75-87.• Luiting, P. and Urff, E.M., 1991. Optimization of a model to estimate residual feed consumption in the laying hen. Livest. Prod. Sci. 27:321-338.• Luxford, B.G. and Beilharz, R.G., 1990. Selection response for litter size at birth and litter size at weaning in the first parity of mice. Theor. Appl. Genet. 80:625-630.• Mackenzie, D.D.S. and Revell, D.K., 1998. Genetic influences on milk quantity. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 97-112.• Makkink, C.A. and Schrama, J.W., 1998. Thermal requirements of the lactating sow. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 271-283.• Massoud, M., Attal, J., Thepot, D., Pointu, H., Stinnakre, M.G., Theron, M.C., Lopez, C., Houdebine, L.M., 1996. The deleterious effects of human erythropoietin gene driven by the rabbit whey acidic protein gene promoter in transgenic rabbits. Repr. Nutr. Dev. 36:555-563.• Monteiro, L.S. and Falconer, D.S., 1966. Compensatory growth and sexual maturity in mice. Anim. Prod. 8:179-192.• Mullan, B.P. and Williams, I.H., 1989. The effect of body reserves at farrowing on the reproductive performance of first-litter sows. Anim. Prod. 48:449-457.• Noblet, J., Etienne, M., Dourmad, J.Y., 1998. Energy efficiency of milk production. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 113-130.• Pérez Enciso, M. and Bidanel, J.P., 1997. Selection for litter size components: a critical review. Genet. Sel. Evol. 29:483-496.• Pluske, J.R. and Dong, G.Z., 1998. Factors influencing the utilisation of colostrum and milk. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 45-70.• Prunier, A., Quesnel, H., Messias de Bragança, M., Kermabon, A.Y., 1996. Environmental and seasonal influences on the return-to-oestrus after weaning in primiparous sows: a review. Livest. Prod. Sci. 45:103-110.• Pursel, V.G., Pinkert, C.A., Miller, K.F., Bolt, D.J., Campbell, R.G., Palmiter, R.D., Brinster, R.L., Hammer, R.E., 1989. Genetic engineering of livestock. Science 244:1281-1288.• Rogowitz, G.L., 1998. Limits to milk flow and energy allocation during lactation of the Hispid Cotton rat (Sigmodon hispidus). Phys. Zool. 71:312-320.
    • 8. General discussion 131• Speakman, J.R. and McQueenie, J., 1996. Limits to sustained metabolic rate: the link between food intake, basal metabolic rate, and morphology in reproducing mice, Mus musculus. Phys. Zool. 69:746-769.• Svendsen, J., 1992. Perinatal mortality in pigs. Anim. Repr. Sci. 28:59-67.• Taylor, St C.S., 1980. Live-weight growth from embryo to adult in domesticated mammals. Anim. Prod. 31:223-235.• Taylor, St C.S. and Murray, J.I., 1987. Genetic aspects of mammalian growth and survival in relation to body size. Academic Press, London.• Taylor, C.R. and Weibel, E.R., 1981. Design of the mammalian respiratory system. Respir. Phys. 44:1-164.• Ten Napel, J., 1996. Genetic aspects of intervals from weaning to estrus in swine. PhD thesis Wageningen Agricultural University, Wageningen, The Netherlands.• Tholen, E., Bunter, K.L., Hermesch, S., Graser, H-U., 1996. The genetic foundation of fitness and reproduction traits in Australian pig populations. 1. Genetic parameters for weaning to conception interval, farrowing interval, and stayability. Aust. J. Agric. Res. 47:1261-1274.• Tomas, F.M., Pym, R.A., Johnson, R.J., 1991. Muscle protein turnover in chickens selected for increased growth rate, food consumption or efficiency of food utilisation: effects of genotype and relationship to plasma IGF-I and growth hormone. Br. Poultry Sci. 32:363-376.• Vangen, O., 1993. Results from 40 generations of divergent selection for litter size in mice. Livest. Prod. Sci. 37:197-211.• Weibel, E.R., 1998. How good is best? Some afterthoughts on symmorphosis and optimization. In: Principles of animal design. The optimization and symmorphosis debate. Weibel, E.R., Taylor, C.R., Bolis, L (ed.). Cambridge University Press, p. 299-306.• Williams, I.H., 1998. Nutritional effects during lactation and during the interval from weaning to oestrus. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 159- 181.• Whithers, P.C., 1992. Comparative Animal Physiology. Saunders College Publishing.• Whittemore, C.T., 1994. Causes and consequences of change in the mature size of the domestic pig. Outl. on Agr. 23:55-59.• Whittemore, C.T., 1996. Nutrition reproduction interactions in primiparous sows. Livest. Prod. Sci. 46:65-83.• Whittemore, C.T., 1998. Influence of pregnancy feeding on lactation performance. In: The lactating sow. M.W.A. Verstegen, P.J. Moughan, J.W. Schrama (ed.). p. 183-200.• Yang, H., Eastham, P.R., Phillips, P., Whittemore, C.T., 1989. Reproductive performance, body weight and body condition of breeding sows with differing body fatness at parturition, differing nutrition during lactation, and differing litter size. Anim. Prod. 48:181-201.
    • 9 RECOMMENDATIONS FOR FURTHER RESEARCH1. Relationship between resource balance and susceptibility to disease and stressAnimals that originate from a population that has been artificially selected for highlevels of production are more at risk for metabolical, reproduction and health problems(Chapter 2). Beilharz et al. (1993) suggest that this may result from an unbalancedresource situation. When genetically forced to high production, disproportionally manyresources (food, body reserves) may be allocated to the production trait, leaving theanimal lacking in ability to respond adequately to other demands (Dunnington, 1990). The present study has investigated changes in food resource allocation patternswith long-term selection for litter size using a mouse model. The results indicate thatnon-reproductive adult female mice selected for high litter size (S-line) allocate alower amount of food resources to maintenance and have therefore more resourcesleft for other resource demanding processes than female mice of a non-selectedcontrol line (C-line) (Chapter 4). However, during lactation, these increased resources,even when combined with a considerable increase in food intake, were not sufficient tosustain the highly increased resource demand without compromising other resourcedemanding processes (Chapter 7). This suggests that these animals, while supportingthe genetically highly increased litter size, may be more at risk for behavioural,physiological and immunological problems. While the present study has concentrated mainly on investigating changes in theresource allocation patterns with selection for litter size, further research should bedirected to investigate how these changes actually relate to susceptibility to diseasesand stress. For this purpose, immune response to diseases, to vaccination or toimmunisation with non-pathogenic substances such as sheep erythrocytes (SRBC) couldbe related to measurements on resource allocation patterns. Furthermore, resourceallocation patterns could be related to physiological stress-response parameters andresults can be discussed in the context of animal welfare.2. Modelling of resource allocation patternsModelling of resource allocation patterns is needed if related parameters are to betaken into consideration in breeding decisions. Such models may be based on severalassumptions. The theory of resource allocation considers that an individual cannotmaximise all its functions with a limited (and insufficient) amount of resources (DeLaguérie et al., 1993). Therefore, correlations between resource consuming functionsare expected to be negative; resources used by one process are no longer available to
    • 134 Generalanother process and thus the increase in one function will compromise another function(Beilharz et al., 1993; De Laguérie et al., 1993). Beilharz et al. (1993) described thistheory with the following equation:R = {RA + RB + RC} + ΣRXi, (1)where R = total amount of available resources, RA, RB and RC are fitness components(e.g., litter size, survival of progeny; for simplicity only three components areconsidered), {RA + RB + RC} = resources used for fitness, and ΣRXi = resources used forprocesses other than fitness. The difference between fitness and other processesis quite arbitrary, since functions that compete for resources may be considered toinfluence fitness in one way or another. Equation (1) can be rewritten as (after Luitinget al., 1997):R = Σ(kM x M) + Σ(kP x P) + Σ(kQi x Qi), (2)where R = total amount of available resources, k = resource conversion factor,Σ(kM x M) = resources used for maintenance, Σ(kP x P) = resources used for productionand Σ(kQi x Qi) = resources used for processes other than maintenance and production.Equation (2) allows for comparison with the equation that is used for the estimation ofresidual food intake (Chapter 6) and includes the fact that the resources consumed bya function depend on the size of the function and on its efficiency in the use of theresources. Tilman (1982) proposed three possible patterns for the response of aprocess to the availability of resources: (1) a linear response curve with constantefficiency of resource transformation until the process reaches its genetic maximumperformance, (2) a negative exponential response curve with decreasing efficiency withincreasing resource level, and (3) a sigmoid response curve which models necessaryfixed costs before launching a response. Both equations (1) and (2) assume thatresources are fully interchangeable between the different functions, i.e., resourcesthat are not used by one process can be used by another process. Furthermore, anappropriate model should incorporate threshold values for several individual processesor combinations of processes. The resource allocation strategies that are realised in a natural population are aresult of the action of the environment in which the population evolves (De Laguérie etal., 1993). Such strategies may be redefined with artificial selection (Chapter 2 andChapter 8).3. Protein turnoverThe metabolic basis for the inverse relationship between high levels of production andhealth and welfare of high producing animals can be investigated by studying theprotein turnover rate. Protein turnover is defined as the continuous breakdown andreplacement of cellular proteins. To measure protein metabolism, one must measure theseparate processes of protein synthesis and protein breakdown, which may varysubstantially either in the same or in opposite directions (Hawkins and Day, 1996). In
    • 9. Recommendations for further research 135growing animals, the rate of turnover equals the rate of protein breakdown, whereas inanimals that lose weight (body protein loss) turnover rate equals the rate of proteinsynthesis; when net body protein balance is zero, turnover rate = rate of proteinsynthesis = rate of protein breakdown (Knap and Schrama, 1996). Protein turnover isessential for life since it provides the flux that is necessary for metabolic regulationand adaptation. It enables the metabolic adjustments that are required forreproduction and development, the repair of damaged tissue, combating infection, andduring or following changes either in the environment or in the nutritional/physiologicalstatus (Hawkins, 1991). Furthermore, as suggested by Pirlet and Arthur-Goettig (1999),specific degradation of functionally defective, old, damaged, denatured proteinmolecules forces the selection of structurally superior proteins. Environmentalinfluences lead to strong and relatively long-lasting conformational changes in theprotein molecules, creating a reservoir of genetic diversity upon which selection canact. Recent evidence shows that conformational changes in the protein moleculesinduced by temperature stress or inactivation of the protective heat shock proteinsaffect nearly any adult morphological structure in fruit flies (Rutherford and Lindquist,1998). Without selection at the level of the proteins, evolution would be impossible(Pirlet and Arthur-Goettig, 1999). Protein turnover is essential for life but also expensive. An average humansynthesises around 5 x 1017 protein molecules per second (Pirlet and Arthur-Goettig,1999). The cells of the body that are never replaced (i.e., more then half of the cells ofthe body (Pirlet and Arthur-Goettig, 1999)) exchange their proteins over 10.000 timeswithin their lifetime (De Duve, 1995). According to Lehninger et al. (1993), of theenergy expended by all cells, 50 to 90% is required for protein turnover. Because ofthe positive relation between protein synthesis and metabolic rate, protein turnoverrepresents a general index of the energy requirements for maintenance, with a minimalestimate of about 20% for the contribution of protein synthesis to maintenancemetabolic expenditure (Hawkins et al., 1989). Therefore, when the body economises onprotein turnover rate, a considerable amount of energy may become available for otherprocesses (Hawkins, 1991). Or in other words, when in a limited resource situation moreresources are allocated towards high production values, the body may be economising onthe level of protein turnover. This hypothesis is supported by results from, e.g., Bates and Millward (1981),Hayashi et al. (1985), Jones et al. (1986) and Tomas et al. (1991), who reported thatbreeding programmes have resulted in slower protein turnover in the isolated tissues ofrats and chickens selected for increased growth rate, leading to lower energyexpenditure and higher growth efficiencies. The lowered costs of maintenance of fastgrowing strains of quails seemed to stem from lower rates of protein breakdown(Maeda et al., 1994). Similar results may be found in animals that are selected for highlitter size. According to Bauman and Currie (1980), in nature, the functions ofpregnancy and lactation have a very high priority over other functions, which allowsthem to proceed at the expense of other metabolic processes even to the point that adisease state is created. In case net protein production for lactation has highestpriority, the body may have to economise on the level of protein turnover due to ashortage of amino acids. Indeed, Roberts and Coward (1984) indicated that lactating
    • 136 Generalrats used significantly less energy for activity and maintenance than virgin animals andsuggested that this is due in part to decreased rates of protein turnover. The energyused for milk production was excluded in this analysis to account for the obligatoryincreases in energy expenditure and metabolism that occur as a result of lactation.With a slower protein turnover rate the space for alertness and adaptability of thebody to environmental stresses and challenges becomes narrower putting the animal ata higher risk to behavioural, physiological and immunological problems. Further research should investigate how changes in resource allocation patternsrelate to changes in protein turnover rates. Insight into this topic may provide solutionsto the problem of production stress in highly producing animals. Metabolic stress maybe overcome by compensating the shortage of aminogenic nutrients by supplementationof the standard ration by amino acids. It can be expected that animals withstand ahigher production intensity after positive intervention. Indeed, results by Fisher et al.(1964) suggest that a protein intake in excess of that required for normal growth ofchickens can be utilised to respond to certain types of stress such as certain aminoacid imbalances and Newcastle Disease virus. When a good understanding of metabolicconstraints for the physiological utilisation of dietary amino acids has been established,a comparison can be made between situations in which the same physiological conditionsare supported by different protein diets.4. Recommendations for further researchSummarised, the following recommendations for further research follow from thepresent study:1. To determine how changes in the resource allocation patterns with selection for litter size actually relate to susceptibility to diseases and stress. Immune response to diseases, to vaccination or to immunisation with non-pathogenic substances should be related to measurements of resource allocation patterns. Resource allocation patterns should be related to physiological stress-response parameters and results should be discussed in the context of animal welfare.2. To model resource allocation patterns so that they can be taken into consideration in breeding decisions. The model may assume (1) limited resource availability, (2) additivity of resources consumed by different processes, (3) interchangeability of resources between different processes, (4) different response patterns of a process to the availability of resources, and (5) threshold values for several individual processes or combinations of processes.3. To determine how changes in resource allocation patterns relate to changes in protein turnover rates. Metabolic stress may be overcome by compensating the shortage of aminogenic nutrients by supplementation of the standard ration by amino acids. When a good understanding of metabolic constraints for the physiological utilisation of dietary amino acids has been established, a comparison can be made between situations in which the same physiological conditions are supported by different protein diets.
    • 9. Recommendations for further research 137References• Bates, P.C. and Millward, D.J., 1981. Characteristics of skeletal muscle growth and protein turnover in a fast-growing rat strain. Br. J. Nutr. 46:7-13.• Bauman, D.E. and Currie, W.B., 1980. Partitioning of nutrients during pregnancy and lactation: a review of mechanisms involving homeostasis and homeorhesis. J. Dairy Sci. 63:1514-1529.• Beilharz, R.G., Luxford, B.G., Wilkinson, J.L., 1993. Quantitative genetics and evolution: is our understanding of genetics sufficient to explain evolution? J. Anim. Br. Gen. 110:161-170.• De Duve, 1995. Vital Dust. Life as a cosmic imperative. New York: Scientific American books, Inc., W. H. Freeman and Company.• De Laguérie, P., Olivieri, I., Gouyon, P.H., 1993. Environmental effects on fitness-sets shape and evolutionarily stable strategies. J. Theor. Biol. 163:113-125.• Dunnington, E.A., 1990. Selection and homeostasis. Proc. 4th World Congr. Gen. Appl. Livest. Prod., Edinburgh, UK (XVI):5-12.• Fisher, H., Grun, J., Shapiro, R., Ashley, J., 1964. Protein reserves: evidence for their utilisation under nutritional and disease stress conditions. J. Nutr. 83:165-170.• Hawkins, A.J.S., 1991. Protein turnover: a functional appraisal. Funct. Ecol. 5:222-233.• Hawkins, A.J.S. and Day, A.J., 1996. The metabolic basis of genetic differences in growth efficiency among marine animals. J. Exp. Marine Biol. and Ecol. 203:93-115.• Hawkins, A.J.S., Widdows, J., Bayne, B.L., 1989. The relevance of whole-body protein metabolism to measured costs of maintenance and growth in Mytilus edulis. Phys. Zool. 62:745-763.• Hayashi, K., Tomita, Y., Maeda, Y., Shinagawa, Y., Inoue, K., Hashizume, T., 1985. The rate of degradation of myofibrillar proteins of skeletal muscle in broiler and layer chickens estimated by N tau-methylhistidine in excreta. Br. J. Nutr. 54:157-163.• Jones, S.J., Aberle, E.D., Judge, M.D., 1986. Skeletal muscle protein turnover in broiler and layer chicks. J. Anim. Sci. 62:1576-1583.• Knap, P. W. and Schrama, J.W., 1996. Simulation of growth in pigs: approximation of protein turn-over parameters. Anim. Sci. 63:533-547.• Lehninger, A., Nelson, D.L., Cox, M.M., 1993. Principles of biochemistry. New York: Worth publishers.• Luiting, P., Vangen, O., Rauw, W.M., Knap, P.W., Beilharz, R.G., 1997. Physiological consequences of selection for growth. 48th Annual Meeting of the EAAP, Vienna, Austria.• Maeda, Y., Kawabe, K., Okamoto, S., Hashiguchi, T., 1994. Comparison of energy metabolism during the growing period in quail lines selected for body weight. Br. Poultry Sci. 35:135-144.• Pirlet, K. And Arthur-Goettig, A., 1999. Maintaining life and health by natural selection of protein molecules. J. Theor. Biol. 201:75-85.• Roberts, S.B. and Coward, W.A., 1984. Lactation increases the efficiency of energy utilisation in rats. J. Nutr. 114:2193-2200.• Rutherford, S.L. and Lindquist, S., 1998. Hsp90 as a capacitor for morphological evolution. Nature 396:336-342.• Tilman, D., 1982. Resource competition and community structure. Princeton, NJ: Princeton University Press.• Tomas, F.M., Pym, R.A., Johnson, R.J., 1991. Muscle protein turnover in chickens selected for increased growth rate, food consumption or efficiency of food utilisation: effects of genotype and relationship to plasma IGF-I and growth hormone. Br. Poultry Sci. 32:363-376.
    • SummaryGenetic selection in livestock species for increased levels of production is oftencompromised by behavioural, physiological and immunological problems (Chapter 2). Thismay result from an unbalanced resource situation. When ‘genetically forced’ to producehighly, disproportionally many resources (food, body reserves) may be (re)allocated tothe production trait, leaving the animal lacking in ability to respond to other demands. The present study investigated changes in food resource allocation patterns withlong-term selection for litter size using a mouse model. It was hypothesised thatanimals selected for high litter size allocate a higher amount of resources to the traitselected for, leaving less resources to respond to other demands. The mice used in thisstudy originated from two lines of the Norwegian mouse selection experiment: a lineselected for more than 90 generations for high litter size at birth (S-line; about 20pups born per litter) and an unselected control line (C-line; about 10 pups born perlitter). Food resource allocation patterns in non-reproductive males and females wereinvestigated in Chapters 3 to 6. In Chapter 3, growth and food intake curves werefitted to individual data. Results showed that the estimates of mature body weight andmature daily food intake were higher in the S-line than in the C-line and higher in malesthan in females. S-line males matured faster than S-line females and C-line mice.Differences in growth and food intake curves can mostly be explained by differences inmature size. Therefore, curve parameters were scaled by individual estimates ofmature body weight. The results showed that scaled mature food intake was higher inthe S-line than in the C-line and higher in females than in males. Scaled maturation ratewas higher in the S-line than in the C-line and higher in S-line males than in S-linefemales. The differences that remained after scaling are a consequence of ‘specificgenetic factors’. The presence of these factors indicates that selection for litter sizehas changed the resource allocation pattern in non-reproductive animals. The existence of specific genetic factors is indicated by variation in efficiencyparameters such as growth efficiency and maintenance requirements. In Chapter 4,estimates of residual food intake and Parks’ estimates of growth efficiency andmaintenance requirements were used to quantify these factors. Residual food intake isdefined as the difference between the food that actually is consumed by an animal andits predicted consumption from observed body weight and growth of the C-linepopulation. Residual food intake is suggested to be an estimate of the amount of‘buffer’ resources that are available for, e.g., physical activity and the ability to copewith unexpected stresses (Chapter 6). Parks’ method involves estimation of growthefficiency and maintenance requirements by fitting curves of body weight againstcumulative food intake. The results showed that at maturity, residual food intake washigher in the S-line than in the C-line and higher in S-line females than in S-line males.Estimates of growth efficiency were higher in the S-line than in the C-line and higher inmales than in females. Estimates of maintenance requirements were higher in the S-linethan in the C-line and higher in females than in males. The higher residual food intake in
    • 140 Generalnon-reproductive adult S-line females indicates that these animals have more resourcesleft to respond to unexpected stresses and challenges. Since S-line females have to support a litter size that has been highly increased byartificial selection, in Chapter 5 it was investigated whether body composition atmaturity (65 days of age) has been affected as a correlated effect of selection forhigh litter size, to sufficiently support the offspring during pregnancy and lactation.Furthermore, part of the observed differences between individuals in residual foodintake may be attributable to differing proportions of body protein and lipid. It wastherefore also investigated if the differences between non-reproductive adult malesand females of the C-line and the S-line in body composition were consistent withexpectations from the differences documented between the lines in RFI (Chapter 4).Relative lipid mass was similar for C-line mice and S-line females; S-line males had alower relative lipid mass. Males had a higher relative protein mass than females, inparticular S-line males. The results showed that body composition in adult non-reproductive females has not been affected as a correlated effect of selection forhigh litter size. Furthermore, the results suggest that the high lean content in S-linemales may explain part of the high residual food intake compared with C-line animalsbut body composition in S-line females was not consistent with the very high residualfood intake documented in these animals (Chapter 4). This means that factors otherthan protein and lipid levels must be responsible for the differences found between thelines and sexes in residual food intake. Several studies have indicated that a higher residual food intake is related to ahigher activity level. Differences in activity may suggest underlying differences incoping strategies in response to unexpected stresses. Therefore, in Chapter 6, it wasinvestigated whether coping strategies in non-reproductive females have been affectedas a correlated effect of selection for increased litter size. Females of the C- andS-lines were subjected to three non-social tests and a social confrontation test. Thenovelty response of S-line females was more dominated by an active coping style thanthat of C-line females. This suggests that the higher residual food intake in S-linefemales compared with C-line females is indeed related to higher levels of activity. The increased amount of food resources that are available to non-reproductiveS-line females at maturity for processes other than maintenance and growth (highresidual food intake; Chapter 4) may be intended for the highly increased resourcedemanding processes of pregnancy and lactation. Food resource allocation patterns inlactating females and the consequences for pup development were investigated inChapter 7. To manipulate experimentally the energy burden of lactation, in each line,half of the females supported litters that were standardised at birth to eight pups perlitter (when larger than eight pups), and half of the females supported the naturallitter size. As for non-reproductive males and females, food resource allocationpatterns in lactating females were quantified by estimates of residual food intake. Theconsequences of food resource allocation patterns for offspring development wereindicated by pre-weaning mortality rates and degree of maturity of the pups from birthto weaning. Residual food intake during lactation was lower in S-line females with non-standardised litters than in C-line females with non-standardised litters. This indicatesthat S-line females that support litters of the size attained by artificial selection
    • Summary 141allocate more resources to the processes that support milk production and haveconsequently fewer resources left to respond to other demands. Residual food intakeduring lactation was lower in S-line females with non-standardised litters than in S-linefemales with standardised litter. After weaning, residual food intake was higher inS-line females than in C-line females, which suggests that S-line females were able torestore the negative resource situation. Pre-weaning mortality rate was higher in S-linenon-standardised litters than in C-line non-standardised litters. Degree of maturity washigher in C-line pups of non-standardised litters than in S-line pups of non-standardisedlitters and higher in pups of standardised litters than in pups of non-standardisedlitters. The results showed that the higher amount of available resources at maturity(Chapter 4), together with a highly increased food intake during lactation were notsufficient to buffer the highly increased resource demand without compromising otherresource demanding processes. This suggests that these animals, while supporting thegenetically highly increased litter size, may be more at risk for behavioural,physiological and immunological problems. Although S-line females with non-standardised litters allocate a particularly high amount of resources towards lactation,this was insufficient to provide the offspring with an adequate amount of resources,resulting in reduced pup development and increased pre-weaning mortality rates.Summarising, the results of the present study indicate that:1. Selection for high litter size has changed the resource allocation pattern in non- reproductive animals, which is indicated by the presence of ‘specific genetic factors’.2. Non-reproductive adult females of the S-line have a higher residual food intake than females of the C-line. This indicates that these females use less resources for maintenance and growth and have therefore more resources available to respond to unexpected stresses.3. Body composition in adult non-reproductive females (65 days of age) has not been affected as a correlated effect of selection for high litter size in order to support the genetically highly increased litter size during pregnancy and lactation.4. Body composition in adult non-reproductive C-line and S-line females is not consistent with the differences in residual food intake between these lines. Factors other than protein and lipid levels must be responsible.5. The novelty response of S-line females is more dominated by an active coping style than that of C-line females. This suggests that the higher residual food intake in S-line females compared with C-line females is related to higher levels of activity.6. Lactating females of the S-line that support their natural litter size have a lower residual food intake than lactating females of the C-line. This indicates that these females use more resources for litter support and have consequently fewer resources left to respond to other demands. These animals may be more at risk for behavioural, physiological and immunological problems. The higher allocation of food resources to litter support was insufficient to provide the offspring with an adequate amount of resources, resulting in reduced pup development and increased pre-weaning mortality rates.