Copyright is owned by the Author of the thesis. Permission is given for a copy to be downloaded by an individual for the purpose of research and private study only. The thesis may not be reproduced elsewhere without the permission of the Author. Maternal Investment in Kaimanawa Horses A thesis presented m partial fulfilment of the requirements for the degree of Doctor of Philosophy In Ecology at Massey University Elissa Zanna Cameron 1998 2 3 "THINK, WHEN WE TALK OF HORSES, THAT YOU SEE THEM PRINTING THEIR PROUD HOOFS I ' THE RECEIVING EARTH" Shakespeare, circa 1599, 'Henry V'. 4 Photo caption: C-band in August 1994. Band stallion, Charly (085), is in the foreground. Mares Celia (132) and Cashew (112) are also visible. Photo by Wayne Linklater ------------------------ - ABSTRACT 5 Maternal investment (MI) was studied in Kaimanawa horses, a population of feral horses in central North Island, New Zealand. It is predicted that individual mares will vary their maternal behaviour so as to maximise their own and their offspring's reproductive success. Maternal investment is defined and measured by maternal input, proximal maternal costs and reproductive costs to the mother. The primary maternal input is milk. Time spent sucking is frequently used to measure milk intake based on the assumption that more sucking equates with proportionately more milk intake. A review found little support for this assumption and so I tested whether sucking time predicted milk intake by labelling the milk of thoroughbred mares with tritium and measuring its transfer to foals . No significant predictive relationship was found. Therefore sucking time cannot be used as an index of milk intake and conclusions about differential MI based on time spent sucking may be wrong. A mare's social environment is a significant modifier of her MI. Mares were more protective, and suffered reproductive costs, in bands with more than one stallion or in other circumstances where paternity is uncertain or rates of aggression are high. Despite individual differences in maternal style, mare behaviour was modified by experimental manipulations of the number of stallions in a band. In an unusual event when a mare and her adult daughter lived in the same band, both co-operated in the care of a single foal. Mare age and experience modified maternal behaviour. As mares age they become more successful at raising foals through better-targeted input, but no extra investment. I tested the Trivers-Willard hypothesis (TWH) of sex differential MI. The hypothesis predicts that mothers in better condition should produce more sons, and invest more in their sons, whereas mothers in poorer condition should favour their daughters. I argue that horses are an ideal species on which to test the TWH. We found no sex biased MI on a population level in terms of maternal input, proximal costs or ultimate costs. However, the TWH makes specific predictions about individuals, not populations. Individual mares in poor condition gave birth to more daughters and biased their MI towards daughters. Conversely, mothers in good condition gave birth to more sons and biased their investment towards sons, supporting all the predictions of the TWH. Mares alter their maternal investment in ways that conform to predictions based on maxirnisation of lifetime reproductive success. ----- ---- --------- �-- - ACKNOWLEDGEMENTS Each chapter has its own acknowledgements section. I thank here those people whose assistance was not specific to any one part. 7 My research project benefited from the input of my supervisors, Edward Minot, Kevin Stafford and Clare Veltman for which I am grateful. Ed Minot, Kevin Stafford and Wayne Linklater read drafts of every chapter and read the thesis in its entirety, thereby enhancing it immeasurably. The research would not have been possible without the support and collaboration of two government agencies: the New Zealand Department of Conservation and the New Zealand Army. The Department of Conservation funded the research through contract 1 850 to Massey University. Liaison was through Bill Fleury at DoC Wanganui. The Army Training Group, Waiouru, gave permission for us to work on their land and provided some logistic support. In particular I thank; Mr John Akurangi, Major Chris Lawrence, Major Neil Bleasdale, Major Bob Campbell, Captain Phil Hughes, Staff Sergeant Jarnie Jones, all from Operations Branch, Headquarters, Army Training Group, Waiouru; Mr Eru Brown, Waiouru Support Company, 4th Logistics Battalion, Army Training Group, Waiouru; Mr John Mangos, Army Training Group Property Management Section, Army Training Group, Waiouru. The Army ' s logistic support, particularly accommodation throughout the study and safety backup, was invaluable. The friendly and helpful day-to-day contact with the Army Training Group, and in particular John Akurangi, Eru Brown and Chris Lawrence, was much appreciated. John Tulloch and his mustering team mustered our horses for two bouts of branding and blood sampling. They also released branded horses from subsequent musters of horses for removal from the range with great tolerance! Jeff Grimmett, Nigel Perkins and Kevin Stafford did the branding and blood sampling, which made an incredible difference to the project. We were lent a crush by Racewell Industries Limited which made branding and blood sampling possible. Keith Henderson at AgResearch Wallaceville and Ian Anderson at the Equine Bloodtyping Centre, Massey University, generously analysed dung samples for pregnancy and blood samples for paternity respectively. Trevor Austin, Chris Lawrence, Peg Loague and Kevin Stafford were instrumental in the stallion removal experiment. Thanks to you all. Alison Franklin introduced us to the horses, and named some of them (yep, Quin is still out there everyone!). Wayne Linklater also named some, though I'm afraid I have to take credit for most of the names. I still think Mr Blobby is a wonderful name! Although field trips were usually spent just with Wayne Linklater, the dogs CBJ and Uzuri), the horses, and the various other inhabitants of the range, I was lucky to have had, at various times, assistance and company in the field from Tarmo P6ldmaa, Rachel Standish, Nokome Bentley, Jenny Lee, Peter "little-cup" Ritchie, Jay McCartney, Kim Carter, Alastair Robertson and Simon Pearce. Jay and Kim were not even put off by the - -- --- ---- - -- 8 erupting volcano (Mount Ruapehu) and an ash fall during field work! My constant field companion was Uzuri, who found dead horses, kept me company, kept me warm, reminded me to eat and told me in no uncertain terms when the day had been too long. The Kaimanawa Wild Horse Preservation always showed an interest in the work that was being done, and it was a pleasure to meet with them, particularly at their field trips onto the horse range. The environment in the Ecology Department (now Ecology Group) provided me with much support. Barbara Just kept finances running smoothly, Jens Jorgenson made some wonderful equipment for the study, and all the rest of the technical staff and especially the secretaries, Erica Reid, Jodi Matenga, Petra van Kan and latterly Rachel Wilson, were always helpful and friendly and made working in the Department a pleasure. Head of Department Dave Lambert was a great student advocate and I thank him and Ed Minot for their support. I got a lot out of BEERS and the Annual Student Colloquium - thanks to all those who contributed. I was also lucky to attend several Conferences during my time at Massey, thanks to the financial support of the Graduate Research Fund and the Ecology Department Development Fund. Thanks to friends, family and colleagues who have discussed ideas or lent me support over the last four years. In particular, thanks to (alphabetical) Julie Alley, Phil Battley, Nokome Bentley, Grant Blackwell, Simon Bulman, Chris Cameron, Halema Flannagan, Carol & Les Linklater, lan McLean, Sharlene Mehrtens, Tarmo P5ldmaa, Peter Ritchie, Alastair Robertson, Rachel Standish and Sara Treadgold. The project benefited immeasurably from the support and assistance of Wayne Linklater. In particular, field work would have been much more difficult alone, and his feedback and commitment are evident throughout the thesis. He even managed to keep me sane throughout - well, almost. For this and much more I am gratefuL My parents, Ian and Zanna Cameron, not only contributed to the project by providing financial security, but are in many ways responsible for the whole thesis. They instilled in me, from a young age, an enthusiasm and respect of animals, an inquiring approach to their lives, and a love of natural history. I thank them for all their help and support throughout the years. Finally, no research is possible without the subjects, and Kaimanawa horses are wonderful to work with. Even those sub-zero days in wind, rain or snow were filled with enjoyment when working with Kaimanawa horses. On a fine clear day, with green grass, fat horses and young foals in the valley, and with an erupting mountain as a backdrop it is just magic. And it doesn' t get much better than that. Title page Frontispiece Abstract --------- --- --- - TABLE OF CONTENTS Acknowledgements Note on the text Introduction Suckling behaviour and milk intake 9 1 3 5 7 10 11 Chapter 1 Is suckling behaviour a reliable predictor of milk 21 intake? A review Chapter 2 Is suckling behaviour a reliable predictor of milk 49 intake? A test Social environment Chapter 3 The influence of social band type on maternal 63 behaviour Chapter 4 Shared offspring care by a mother and her 85 daughter Mare age and experience Chapter 5 Aging, experience and increasing reproductive 97 success Offspring sex Chapter 6 Birth sex ratios relate to mare condition at 111 conception Chapter 7 Sex differential maternal investment Discussion 123 143 1 0 NOTE ON THE TEXT Each chapter is set out largely in the style of the journal to which it has been submitted. Consequently, there is some repetition, particularly in methods sections, and there are stylistic differences between chapters. In addition, other authors are included in the paper reference. For each chapter, my input was the greatest. I planned the research, undertook the field work, analysed the data and wrote the manuscripts . I was, however, helped by my co-authors. Kevin Stafford, Edward Minot and Clare Veltman were my supervisors. Wayne Linklater was studying the Kaimanawa horses for his own PhD and thereby contributed to most aspects of my study. All research described in my thesis was approved by the Massey University Animal Ethics Committee. - - ----- 1 1 Introduction H • • • INDIVIDUALS WHICH GENERATED OR NOURISHED THEIR OFFSPRING BEST WOULD LEA VE ... THE GREATEST NUMBER" Darwin, 1871, 'The Descent of Man and Selection in Relation to Sex' 1 2 Photo caption: Black Bess (074) pregnant in 1996. Photo by Wayne Linklater -- ------- 1 3 The maternal behaviour of an individual animal may differ from others of the same species . It is predicted that mothers will invest in their offspring so as to increase their own reproductive success and improve their inclusive fitness through increased reproductive fitness of their offspring (Trivers 1 972, Trivers & Willard 1 973, Maynard Smith 1 980, Clutton-Brock 1 99 1 ) . Maternal investment is defined as "any investment by the parent into the current offspring that increases its chance of surviving (and hence reproductive success) at the cost of the parent' s ability to invest in future offspring" (Trivers 1 972). As such, it incorporates both maternal input, proximal maternal costs and costs to future reproduction (Clutton-Brock 1 99 1 ). Studies frequently do not consider the three components of maternal investment. In studies of maternal investment there are several often cited confounding factors that complicate the interpretation of measures of variation (eg. future competition with philopatric offspring, Clark 1978, Silk 1 983; litter size, Williams 1 979). For example, there are several different ways in which a mother can change her maternal investment if there is more than one offspring per litter, or if she can reproduce more than once in a season. Where she has only one offspring, variation is limited to the form and duration of maternal investment in the single offspring (Williams 1979, Gosling 1 986). Studies on sex differential maternal investment suffer from the influence of confounding variables (Clutton-Brock 1 99 1 ). It is predicted that, in polygynous species, mothers with more to invest should favour sons, as mothers in better condition would be able to raise a successful son who would sire more offspring than a successful daughter who would be limited to a maximum of one offspring per year. Conversely, mothers with less to invest would be less likely to raise a son that would compete well enough to be able to breed, so that a daughter would raise more offspring than a son would. Therefore, these mothers should invest more in daughters (Trivers-Willard Hypothesis, TWH, Trivers & Willard 1973) . Although prima facie the TWH appears to apply best to species that are sexually dimorphic as well as polygynous, sexual dimorphism can confound the relationship. In sexually dimorphic species Maynard Smith ( 1980) showed that it could be an evolutionary stable strategy for mothers to invest more in sons on a population wide basis. However, the Trivers-Willard hypothesis predicts that individual mothers would behave differently toward sons and daughters on the basis of their ability to invest. Therefore studies of polygynous yet sexually monomorphic species with a single offspring are particularly valuable; males do not have obligate faster growth so observed differences will be due to differential maternal investment. In addition, where one sex is philopatric while the other disperses there are added factors that may influence sex differential maternal investment. Philopatric daughters may inherit their mothers rank (Hiraiwa-Hasegawa 1 993), compete with their mothers for mates (Hamilton 1967) or resources (Clark 1 978, Silk 1 987), and any differences in 1 4 investment i n males early i n life may be compensated for by prolonged low levels of investment in daughters (Clutton-Brock 1 99 1 ) . The horse as a model for testing maternal investment theory. Horse populations, which are scattered throughout the world (Linklater 1 998) have been the subject of much research (eg. Tyler 1 972, Feist & McCullough 1976, Gates 1 979, Berger 1 986, Duncan 1 992). However, although maternal behaviour is known to vary between individuals, there has been little research on causes of such variation, and those studies that have looked at differential maternal investment have found conflicting results. For example, sex differential maternal behaviour was described in some studies (Duncan et al. 1 984, Berger 1 986), but not found in others (Crowell-Davis 1 985, 1 986). Horses are ideal for testing theories of maternal investment for several reasons. Confounding variables are at a minimum. Litter size is constrained at one, both sons and daughters disperse from their natal social groups and sexual dimorphism is minimal (Feh 1 990, Willoughby 1 974). Extensive research on domestic and feral horse popuiations (eg. Boyd 1 979, Boy & Duncan 1 980, Carson & Wood-Gush 1 983 , Duncan et al. 1 984, Crowell-Davis 1 985, Berger 1 986, Crowell-Davis 1 986, Crowell-Davis et al. 1 986, Duncan 1 992, Martin et al. 1 992, Monard et al. 1997) shows that the assumptions of hypotheses differential maternal investment apply. In addition, it is easy to validate research methodology used in feral horses on domestic horses. Horses are a large and easily identifiable animal with a clearly defined period of maternal investment, which facilitates the study of maternal investment from conception to offspring dispersal. Maternal behaviour varies widely between individual mothers. My study investigated variation in maternal investment in Kaimanawa horses with a view to answering three questions: Does the social environment influence maternal investment? How does mare age and experience influence maternal behaviour? Do mares change their maternal behaviour in relation to the sex of their offspring? The study population. Feral horses, known locally as Kaimanawa horses, live in the Kaimanawa ranges and their surrounding plateaux (64000 ha) in the central North Island of New Zealand (Figure 1 ) . Their range includes the New Zealand Army Training Area near Waiouru, where they have lived since the late 1 800s (Rogers 1 99 1 ) . The horse population, its range and the study site are described in Linklater ( 1 998), as is the study area and focal population centred in the lower Moawhango river basin. Fifty five mares and their foals born in 1 994, 1995 and 1 996 are the focus of this study and some other mares in the population, and foals born in 1 997 , are used in some analyses. ---------�.� - - Figure 1 . Location of the study site in the North Island of New Zealand. North Island Kaimanawa feral ___ --��� horse range Christchurch South Island 15 1 6 The horses live in stable social groups typically containing one, but sometimes up to four stallions with mares and their offspring. These bands are loyal to a home range that overlaps in part or full with the home ranges of other bands. Both sons and daughters disperse from their natal band, so mares within a band are not closely related. Stallions without bands are found in bachelor groups that have unstable membership. This social structure is similar to that found in all feral horse populations (reviewed in Linklater 1 998), and in plains and mountain zebras (Klingel 1972) . The breeding season extends from September to April. Foals are born after an eleven month gestation, with a peak in births between October and December (Fig 2). Foals are suckled for at least 8 months, but are sometimes suckled for up to 2 years and foals continue to live with their mothers until dispersal at around 1 5 months. Daughters disperse into other bands, whereas sons become bachelors. I studied maternal behaviour from foal conception to dispersal. Figure 2 . Timing of foal births in 1 994, 1 995, 1 996 and 1 997. 0.4 c: I-< 0 0.3 ..D c: .S ....., I-< 0 0.2 0-0 I-< 0... 0. 1 Sept Oct Nov Dec Jan Feb Mar Apr Month The study. Studies on maternal investment often use maternal input as an indication of investment. In mammals the primary input is the resources transferred from mother to offspring in the form of milk. However, it is not usually possible to measure milk intake in the field, and time spent sucking is often used as a predictor of milk intake based on the assumption that offspring who suck more get more milk. As recent studies have cast doubt on this 1 7 assumption, I begin by reviewing the evidence that time spent sucking can reliably predict milk intake (Chapter 1 ) . I then test whether sucking time predicts milk intake in horses (Chapter 2) by labelling the milk of thoroughbred mares with radioactive isotopes, measuring milk transferred to foals, and sampling their behaviour in a simulation of field measurements. In the next section I look at the influence of social environment on maternal behaviour. The risk of infant injury or mortality has a significant modifying effect on the behaviour of mothers in many species. In horses, infanticide has been reported, as has increased accidental death rates of foals due to high rate of stallion aggression. I investigate stallion aggression as a modifier of maternal behaviour, particularly in relation to the presence of more than one stallion in a band (Chapter 3) . In multi-stallion bands there is a higher rate of aggression, and a lower certainty of paternity. Some mares changed band type and others had their multi-stallion bands experimentally reduced to a single stallion band. Therefore, the influence of the additional stallion on maternal behaviour could be measured. I document the first reported case of shared suckling in horses (Chapter 4), and investigate the circumstances in which shared suckling or offspring care do or do not occur, and determine if the mares are co-ordinating and co-operating their foal care. It has been hypothesised that mammalian mothers become more successful at raising offspring as they age due to declining residual reproductive value (Pianka & Parker 1 975, Clutton-Brock 1 984) . An alternative hypothesis suggests that mares become more successful because of greater experience, not because of increased maternal investment (Fairbanks 1996). In Chapter 5 , I test predictions from these two hypotheses to determine if mares do become more successful at rearing foals, and whether the success can be attributed to greater effort by older mares, or to the greater experience of older mares. I then test theories of sex differential maternal investment. In Chapter 6, I outline the assumptions and predictions of the Trivers-Willard hypothesis (TWH) of sex­ differential maternal investment. The TWH predicts that mothers in relatively better condition, and therefore with more resources to invest, would be advantaged by producing more sons, whereas females in poorer condition would be advantaged by producing more daughters. I test if there is any variation in birth sex ratios in relation to maternal body condition. In Chapter 7, I extend the hypothesis, as did Trivers & Willard ( 1 973), to determine if mares in better condition invest more in sons and mares in poorer condition invest more in daughters . Other studies of the TWH use population measures of matemal investment. In both chapters I apply the TWH to individual mares, as the TWH predicts that individual mares adjust their strategy in relation to their particular situation. Finally, I discuss variation in maternal investment in Kaimanawa horses and discuss the implications of these conclusions. 18 REFERENCES Berger, l. 1986. Wild Horses of the Great Basin. Chicago: University of Chicago Press. Boy, V. & Duncan P. 1980. Time-budgets of Camargue horses 1. Developmental changes in the time-budgets of foals. Animal Behaviour 71, 187-202. Carson, K. & Wood-Gush, D . G. M. 1983. Behaviour of thoroughbred foals during nursing. Equine veterinary Joumal 15, 257-262. Clark, A. B . 1978. Sex ratio and local resource competition in a prosimian primate. Science 201, 163-165. Clutton-Brock, T. H . 1984. Reproductive effort and terminal investment in iteroparous animals. American Naturalist 123, 212-229. Clutton-Brock, T . H . 1991. The Evolution of Parental Care. Princeton University Press, Princeton, New Jersey. Crowell-Davis, S . L. 1985. Nursing behaviour and maternal aggression among Welsh ponies (Equus caballus). Applied Animal Behaviour Science 14, 11-25. Crowell-Davis, S . L. 1986. Spatial relations between mares and foals of the Welsh pony (Equus caballus) . Animal Behaviour 34, 1007-1015. Crowell-Davis, S . L. , Houpt, K. A. & Carini, C . M. 1986. Mutual grooming and nearest-neighbour relationships among foals of Equus caballus. Applied Animal Behaviour Science 15, 113-123. Duncan P. 1992. Horses and Grasses. New York: Springer-Verlag. Duncan, P. , Harvey, P. H . & Wells, S . M. 1984. On lactation and associated behaviour in a natural herd of horses. Animal Behaviour 32, 255-263. Feh C, 1990. Long term paternity data in relation to different rank aspects for Camargue stallions. Animal Behaviour 40, 995-996. Feist, 1. D. & McCullough, D. R. 1976. The behaviour patterns and communication in feral horses. Zeitschriftfur Tierpsychologie 41, 337-371. Gates, S . 1979. A study of the home ranges of free-ranging Exmoor ponies. Mammal Review 9, 3-18. Gosling, L. M. 1986. Selective abortion of entire litters in the coypu: adaptive control of offspring production in relation to quality and sex. American Naturalist 127, 772- 795. Hamilton, W. D. 1967. Extraordinary sex ratios. Science 156, 477-488. Hiraiwa-Hasegawa, M. 1993. Skewed birth sex ratios in primates: should high-ranking mothers have sons or daughters? Trends in Ecology and Evolution, 8, 395-400. Klingel, H . 1972. Social behaviour of African Equidae. Zoologica Africana 7, 175-185. Linklater, W. L. 1998. Social and spatial organisation of horses. PhD thesis, Massey University. Maynard-Smith, l. 1980. A new theory of sexual investment. Behavioral Ecology & Sociobiology 7, 247-251. ---- ------- -- --- Pianka, E. R. & Parker, W. S . 1 975. Age-specific reproductive tactics. American Naturalist 1 09, 453-464. Rogers, G. M . 1 99 1 . Kaimanawa feral horses and their environmental impacts. New Zealand Journal of Ecology 15 , 49-64. Silk, 1. B . 1 983. Local resource competition and facultative adjustment of sex ratios in relation to competitive abilities. American Naturalist. 1 2 1 , 56-66. Trivers, R. L. 1 972. Parental investment and sexual selection. In: Sexual Selection and the Descent of Man. (Ed. by B . Campbell) , pp. 1 36- 179. Chicago: Aldine. Trivers, R. L. & Willard, D. 1 973. Natural selection of parental ability to vary the sex ratio of offspring. Science 1 79, 90-92. Tyler, S. 1. 1 972. The behaviour and social organisation of the New Forest ponies. Animal Behaviour Monograph 5, 85-344. Williams, G. C. 1 979. The question of adaptive sex ratio in outcrossed vertebrates. Proceedings of the Royal Society of London B 205, 567-580. Willoughby, D. P. 1 974. The Empire of Equus. New Jersey: A. S. Barnes. 1 9 20 CHAPTER 1 Suckling Behaviour and Milk Intake Is SUCKLING BEHAVIOUR A RELIABLE PREDICTOR OF MILK INTAKE? A REVIEW "THE COW IS OF THE BOVINE ILK, ONE END IS MOO, THE OTHER MILK" 21 Ogden Nash, 1931, 'The Cow ', 22 Photo caption: Kandy-Floss with foal Popcorn in 1986. Photo by Elissa Cameron 23 ABSTRACT In studies on mammalian parental investment, time spent suckling is often used as a predictor of the milk transferred from mother to infant. It is assumed that the rate of milk transfer is positively correlated with the time spent suckling. However, this assumption has not been tested and empirical studies show conflicting results. Nevertheless, in species in which suckling can readily be observed, time spent suckling is still used to measure milk transfer, although an increasing number of workers recognize that the measure is potentially inaccurate. A meta-analysis on studies that have correlated measures of time spent suckling with milk intake estimates based on weight gain, has revealed a weak positive relationship and significant heterogeneity between studies. Isotope label-4ng techniques for the measurement of milk transfer independent of behaviour have been in use since the 1 970s. The technique has been used particularly to study milk transfer in species in which suckling is difficult to observe. Only one study has attempted to correlate behavioural measures with independent isotope measures, and it found no relationship between the two measures. I suggest that researchers have avoided such a test as it is unlikely that a strong relationship will be found between milk transfer and suckling behaviour, and I discuss the various factors that confound the relationship and contribute to high heterogeneity between studies. Consequently, the assumption that milk transfer can be measured by time spent suckling has inadequate empirical foundation, and needs to be tested using isotope labelling methods. C hapter reference: Cameron, E. Z. in press. Is suckling behaviour a reliable predictor of milk intake? A review. Animal Behaviour INTRODUCTION Although parental investment is defined in terms of costs to the parent in future reproductive success (Trivers 1972), it is usually measured as the resources allocated to the offspring, or parental input (Evans 1990; Birgersson & Ekvall 1 994) . In mammals, the most obvious parental inputs are the resources transferred during lactation (Pond 1 977; Lee 1 987), which are more energetically costly than the prenatal costs of gestation (Martin 1 984; Clutton-Brock et al. 1 989). In studies of mammalian parental investment an estimate of milk obtained by offspring is therefore a fundamental measure of parental input (Martin 1 984; Lee 1 987; Kretzmann et al. 1 993). However, the measurement of milk transferred from mother to offspring is difficult in field situations (Kretzmann et al. 1 993), and consequently behavioural measures of suckling have been used to estimate amounts of milk transferred. Such studies assume that all other factors being equal, offspring that suck more obtain more milk (Fletcher 197 1 ; Berger 1 979). 24 The assumption that the time spent suckling is directly proportional to the amount of milk transferred has not been tested (Shackleton & Haywood 1 985; Babbitt & Packard 1 990b) and has recently been questioned (eg. Anderson & Fedak 1 987; Festa-Bianchet 1 988; Mendl & Paul 1 989; Babbitt & Packard 1 990a; Lavigueur & Barrette 1 992; B irgersson & Ekvall 1994). Some initial studies found anomalies that suggested that the relationship between suckling and milk transfer may not be as simple as had been assumed (eg. Ewbank 1964; Fletcher 1 97 1 ; Hall et al. 1 978; Berger 1 979; Loudon & Kay 1984), and subsequent tests of the relationship have failed to show a significant positive correlation (Mend 1 & Paul 1 989; Birgersson & Ekvall 1 994). Consequently, some authors have questioned the reliability and utility of such estimates of milk transfer (eg. Berger 1 979; Shackleton & Haywood 1 985; Babbitt & Packard 1 990 a, b; Alley et al. 1 995 ; Pelabon et al. 1995). In this study, I review the evidence that suckling behaviour is related to milk transfer in placental mammals and perform a meta-analysis on these data. I suggest reasons why the relationship between suckling behaviour and milk intake might be weak and examine the use of isotope labelling techniques to test the assumption that milk transfer can be estimated by time spent suckling. TIME SPENT SUC KLING AND MILK INTAKE A number of studies have correlated time spent suckling with growth rates or weight gain during a suckle bout as indices of milk intake (Table 1 ) . Of the studies that correlated infant growth rates with time spent suckling, only one (5%) found a significant positive relationship between growth and time suckling, three a significantly negative relationship and 1 6 no significant relationship. However, the relationship between growth and suckling can be obscured when the mother' s milk is not the only source of nutrition to the young such as in precocial young that feed on solid food shortly after birth (Martin 1 984). Even where milk is the primary source of nutrition other factors such as differences in activity (Duncan et al. 1984; Byers & Bekoff 1990), metabolic rates (Mendl & Paul 1 989; Kretzmann et al . 1 993), genetic growth potential (Yates et al. 197 1 ) or differences in nutrient assimilation efficiency (Verme 1 989), could lead to differences in growth rate despite ingestion of the same quantity of milk (Clutton-Brock 1 99 1 ). In particular males may be more efficient at the conversion of milk energy into growth (Kretzmann et al. 1 993) as testosterone may enhance growth by allowing greater efficiency of nutrient assimilation (Glucksman 1 98 1 ) . In addition, variation in milk quality and composition both between and within individuals (Oldham & Friggens 1989) could cause marked differences in energy intake even if the amounts of milk ingested were similar (Berger 1 979). Studies that have investigated the relationship between time spent suckling and estimates of milk intake by weighing an offspring before and after a suckle bout ( 'test- _. - 25 Table 1 . Summary of studies that have correlated measures of time spent suckling with estimates of milk intake based on infant weight gain (growth or test-weighing) Measure Species 1"" pb nC Classd Source GROWTH Suckle bout duration reindeer 0.04 NS 7 <45 d Lavigueur & Barrette 1 99 2 human nd NS 45 Buue et al. 1 985 human -0.28 NS 46 de Carvalho et al . 1 98 2 Suckle bout frequency reindeer 0.54 NS 7 <45 d Lavigueur & Barrette 1 99 2 reindeer 0.10 NS 5 46 -1 00 d sheep t'8=2.27 + 1 9 0-3 wks Ewbank 1 967 sheep t'8=0.68 NS 19 3 -6 wks sheep 0.04 NS 34 1 -2 wks Fletcher 1 97 1 sheep 0.20 NS 34 3 -6 wks sheep 0.02 NS 34 7 -1 2wks human nd NS 45 Butte et al . 1 985 human 0.24 NS 46 de Carvalho et al. 1 98 2 Total time suckling reindeer 0.29 NS 7 < 45 d Lavigueur & Barrette 1 99 2 fallow deer 0.30 NS 22 all fawns" Birgersson & Ekvall 1 994 fallow deer 0.06 NS 1 2 all fawns B irgersson & Ekva\l 1 997 fallow deer 0.63 NS 1 2 > 2 wks Soay sheep 0.04 NS 14 1 988 Robertson et al . 1 99 2 Soay sheep -0.64 14 1 989 Scans mouse -0.53 1 6 all l itters Mendl & Paul 1 989 cat -0.43 24 all litters TEST-WEIGHING Suckle bout duration fur seal 0.49 + 73 Tril lmich 1 986 human -0.14 NS 46 de Carvalho et al . 1 98 2 human nd NS 44 de Carvalho et al. 1 983 human nd NS 45 Butte et al. 1 985 human FI.4<)= I. 9 NS 5 2 Drewett et aL 1 989 human 0.13 NS 73 to 3 mths Dewey et al. 1 99 1 Suckle bout frequency human 0.19 NS 46 de Carvalho et al . 1 98 2 human nd + 44 0- 2 wks de Carvalho et al 1983 human nd NS 44 I mth human nd NS 45 Butte et al. 1 9 85 human FI.49=7.8 + 5 2 Drewett et al . 1 989 human 0.58" + 5 2 human 0.41 + 27 Rattigan et al. 1 98 1 hum� 0.13 NS 73 to 3 mths Dewey et al . 1 99 1 red deer nd 1 9 food type Loudon & Kay 1 984 red deer nd 1 7 food type Loudon et al. 1 983 Total time suckling human 0.69" + 5 2 Brown et aL 1 98 2 human 0.57 + 5 2 Drewett et al . 1 989 human 0.47 + 76 Imong et aL 1 989 human 0.24 + 73 to 3 mths Dewey et al . 1 99 1 red deer nd + 17 food type Loudon et al. 1 983 a: Correlation reported by author, nd == no data supplied; correlation values in bold are those used in the meta -analysis. One study (fallow deer fawns> 2 weeks) was excluded because the authors used data that were a subset of data they used to calculate the correlation for all fawns. b: Direction of relationship as reported by author; ns no significant relationship, + significant positive relationship, - significant negative relationship c: Number of individuals in the study d: Specific age or other classification on which the test was performed t Improved to r=0.6 2 when only fawns over 1 4 days with multiparous mothers used * Age of infant taken into account 26 weighing' , eg o Butte et al . 1 984, or 'weigh-suckle-weigh' , eg o Sadleir 1 980) have also shown ambiguous results; 1 0 (48%) found a significant positive relationship, two found a significant negative relationship and nine found no significant relationship (Table 1 ) . In humans test-weighing can provide an accurate measure of milk transfer as the weighing does not disrupt normal suckling patterns (Arthur et al. 1 987). Butte et al. (1984) found that estimates of milk transfer by test-weighing were accurate when compared with estimates from the transfer of radio-isotope labelled milk, although Brown et al. ( 1982) found that test-weighing underestimated milk transfer when measured against known amounts of milk. In animals other than humans, offspring have normally been kept off their mother for a period before the suckle (eg. Loudon et al. 1 984; van der Steen & de Groot 1 992), so that the test-weighing interferes with normal suckling patterns (Oftedal 1 984; Wright & Wolff 1 976; Pettigrew et al. 1 985). In addition, errors associated with calculating a weight differential that is a small proportion of the animal being weighed can be large (Pettigrew et al. 1 985). Consequently, test-weighing procedures tend to provide less accurate estimates of milk intake in animals other than humans. Despite the problems associated with both methods, these provide the best available estimates of the correlation between time spent suckling and milk intake. Consequently , I combined these data and performed a meta-analysis to identify significant trends within the data set. Meta-analyses are a quantitative technique for combining the results of statistical tests on previous studies that have looked at the same relationship, in this case correlations between weight gain and time spent suckling, taking into account the sample size of each study. The goal of the meta-analysis is to determine if there is an overall significant trend in the studies and to examine the heterogeneity of the data set (Hedges & Olkin 1 985). I found that there was a significant positive relationship between all measures of time spent suckling and estimates of milk intake based on weight gain, but that less than 1 0% of the variation was explained by the relationship between the two variables (Table 2). Of the behaviour sampling methods, only suckle bout frequency and total time suckling were significantly correlated with estimated milk intake, with total time suckling providing better estimates. Similarly , estimates of milk intake based on growth were not significantly correlated with suckling time, but test-weighing and suckling time were correlated. Even when we combined the best behaviour estimate, total time suckling, with the best intake estimate, test -weighing, less than 25% of the variation in milk intake was explained by suckling time (p2=0.24, P • -, ro "0 '- '" 1:: . 0.80 ..0 . 6 ..0" V 0 u ..o c "'-"' 4 C';l <;;; :a c: 2 C';l v ::E 0 I I I I I 0-20 2 1 -50 5 1 -1 10 1 1 1 -200 wean Foal age (days) Figure 2 . Spatial relationship of mares and foals in single and multi-stallion bands as foals age: a) proportion of the approaches between mare and foal due to the foal (lower scores therefore indicate more mare effort) , and b) the mean distance between the mare and foal. Circles indicate foals raised in single stallion bands with triangles multi-stallion bands; Lines are sons, dashed lines daughters. 75 single stallion bands (Fig 2a). However, the mean distance between mare and foal did not vary between band types (Fig 2b) . When samples between birth and 1 10 days of age were combined as the period of essential investment for foal survival the MANOV A was significant for daughters (Wilk' s Lambda=0.56, F7J3=3.7, P<0.005) and showed the same trend for sons (Wilk' s Lambda=0.68 , F7,29=2.0, P ", ,,, .0 -0 -0.20 ", c: () c:: � � -0.40 4-CS - - r I I I l I - I dIstance I approach I epIsodes I suck dam to foil by fOal enred attempts by dam unsuccessful 1 1 I epIsodes I per bout Figure 6. The effect of foal loss on subsequent maternal behavior. Bars indicate the difference in 5 maternal behaviors between the foal that died and the subsequent foal . The subsequent foal was closer to its mother, its mother approached it more often, it had fewer unsuccessful sucks and more episodes in each suck bout. 79 which was greater for foals in multi-male bands. In particular, mares in multi-stallion bands were more likely to approach their foals after an approach by a stallion (see Crowell-Davis & Houpt, 1 986 for photo series of similar protectiveness in domestic horses). Mares in multi-stallion bands were more protective toward their foals, and they seemed to be responding to the possible risk of injury to the foal, as predicted by stallion aggression to the foal. Foal rearing was also more costly to mares in multi-stallion bands; they lost more condition and were less likely to foal in the subsequent year. Mares that shifted bands or had their band experimentally reduced by the removal of the subordinate male were more protective when there was more than one male in the band. Social disturbance and the presence of unfamiliar males induces similar protective behaviour in other species (eg. Collins et al. , 1 984; Fairbanks and McGuire, 1 984) and protectiveness reduces the likelihood of infants being killed (Collins et aI. , 1 984) . Therefore mares in single and multi-stallion bands do not have inherently different maternal strategies but adjust their behavior to the proximal potential risk. Each mare had her own maternal style, as has been found previously in a variety of species (eg. Fairbanks, 1986), including horses (Crowell-Davis, 1 986a), but each was more protective in multi-stallion bands. Similar results have been found in primate groups where an unfamiliar male is present (Fairbanks, 1 996) . In addition, within a single-stallion band mares were more protective of foals that were not sired by the band stallion. In horses, reports of stallion-foal aggression have been confined to circumstances where there is uncertainty of paternity (Boyd, 1 99 1 ; Duncan, 1 982; Ryder and Massena, 1 988). Protective behavior is induced in the mothers of many species with the presence of non-paternal males (Fairbanks, 1 997; Fairbanks and McGuire, 1 987; Hrdy, 1 977; Packer and Pusey, 1 983), or other individuals that are a threat to offspring (Maestripieri, 1 995, 1998; Parmigiani et al. , 1 994) . Experimental manipulations of social environments are particularly valuable, as few studies have been able to separate cause and effect (Maestripieri, 1 998). We show that mares are both assessing and responding to their current social environment, particularly differences in stallion behaviour, as well as assessing risk in relation to foal paternity. Mares in multi-stallion bands seemed to have more costs associated with foal rearing, which may be due to the higher rates of aggression in multi-stallion bands that result in extra effort in the form of protective behavior (Linklater, 1 998). Multi-stallion band mares were less likely to foal in the subsequent year, lost more weight during the period of peak lactation, and raised offspring who were in poorer condition as yearlings. Although we found no difference in foal survivors hip in our focal popUlation, Linklater ( 1 998) shows that in the greater population of Kaimanawa mares, multi-stallion band females are more likely to lose their foal from conception to independence than are single stallion mares. In addition, Duncan ( 1 992) suggested that social instability and stallion aggression contributed to high rates of neonatal death during some years . 80 Previous studies have found that mares who fail to form stable consort relationships or change groups regularly have lower reproductive success (Kaseda et al. 1 995, Rutberg and Greenberg 1 990), and suggest that aggressive interactions may interfere with successful reproduction indirectly (Rutberg and Greenberg 1 990), or directly by feticide (Berger 1 983) or infanticide (Duncan 1 982). Linklater ( 1998) showed that mares that were not stable group members behave similarly to mares in multi-stallion bands. Therefore, aggression by stallions, and the potential for infanticide are important factors that modify maternal behavior and influence mare reproductive success. Mares in multi-stallion bands and in other situations where paternity is uncertain expend greater maternal effort in the form of protective behaviors, and consequently have lower reproductive success. A CKNOWLEDGMENTS This study was funded by Department of Conservation contract number 1 850 to Massey University . We thank the New Zealand Army for providing permission to work in the Army Training Area, and for providing some logistic support. We thank Clare Veltman for her input and Ian Anderson and the Massey University Equine Bloodtyping Centre for paternity analysis. REFERENCES Agooramoorthy G, Mohnot SM, Sommer V, Srivastava A, 1 988. Abortion in free ranging Hanuman langurs (Presbytis entellus) - a male induced strategy? Hum Evo1 3 :297-308. Becker CD, Ginsberg JR, 1 990. Mother-infant behavior of wild Grevy's zebra: adaptations for survival in semi-desert East Africa. Anim Behav 40: 1 1 1 1- 1 1 1 8 . Berger J , 1 979. Weaning conflict in desert and mountain bighorn sheep ( Ovis canadensis): An ecological interpretation. Z Tierpsychol 50: 1 88-200. Berger J, 1 983 . Induced abortion and social factors in wild horses. Nature 303 :59-6 1 . Berger J , 1986.Wild Horses of the Great Basin. Chicago; Chicago University Press. Borries C, 1 997. Infanticide in seasonally breeding multimale groups of hanuman langurs (Presbytis entellus) in Ramnagar (South Nepal ) . Beh Ecol Sociobio1 4 1 : 1 39- 1 50. Bowling AT, Touchberry RW, 1 990. Parentage of Great Basin feral horses. J Wildl Manage 54:424-429. Boyd LE, 1 99 1 . The behavior of Przewalski' s horses and its importance to their management. Appl Anim Behav SC! 29: 30 1 -3 1 8. Byers JA, Bekoff M ( 1 992) Inference in social evolution theory: a case study. In: Bekoff M, Jarnieson D (eds) Interpretation and Explanation in the Study of Animal Behavior Vo1 2 Explanation, Evolution and Adaptation. Westview Press, Boulder, pp 84-97 8 1 Cameron EZ, in press. Is suckling behavior a useful predictor of milk intake? A review. Anim Behav Cameron EZ, 1 998. Maternal investment in Kaimanawa horses. PhD thesis, Massey University, New Zealand. Cameron, E. Z., Stafford, K. J . , Linklater, W. L & Veltman, C. J. in press. Suckling behaviour does not measure milk intake in horses. Anim Behav Carroll CL, Huntingdon PJ, 1 988 . Body condition scoring and weight estimation of horses . Equine vet J 20:4 1 -45. Cars on K, Wood-Gush DGM, 1 983 . Behaviour of thoroughbred foals during nursing. Equine vet J 1 5 :257-262. Clutton-Brock TH, 1 99 1 . The Evolution of Parental Care. Princeton: Princeton University Press. CouIon J, Graziani L, Allaine D, Bel MC, Pouderoux S , 1 995. Infanticide in the alpine marmot (Mannota mannota) . Ethol EcoI Evo1 7 : 1 9 1 - 1 94 Collins DA, Busse CD, Goodall J, 1 984. Infanticide in two populations of savanna baboons. In: Infanticide: Comparative and Evolutionary Perspectives (Hausfater G, Hrdy SB eds) . New York: AIdine; 1 93-2 1 5 . Crockett CM, Sekulic R , 1 984. Infanticide i n red howler monkeys (Aloutta seniculus). In: Infanticide: Comparative and Evolutionary Perspectives (Hausfater G, Hrdy SB eds). New York: AIdine; 1 73- 1 9 1 Crowell-Davis SL, 1985 . Nursing behavior and maternal aggression among Welsh ponies (£quus caballus) . Appl Anim Behav Sci 14: 1 1 -25. Crowell-Davis SL, 1 986. Spatial relations between mares and foals of the Welsh pony, £quus caballus. Anim Behav 34: 1007- 10 15 . Crowell-Davis SL, Houpt KA, 1 986. Maternal behavior. Vet Clinics Nth Am: Equine Practice 2 1 :557-57 1 . Duncan P, 1 982. Foal killing by stallions. Appl Anim Etho1 8 :567-570. Duncan P, 1 992. Horses and Grasses. New York: Springer-Verlag. Duncan P, Harvey PH, Wells SM, 1 984. On lactation and associated behavior in a natural herd of horses. Anim Behav 32:255-263. Eagle TC, Asa C, Garrott RA, Plotka ED, Siniff DB, Tester JR, 1 993. Efficacy of dominant male sterilisation to reduce reproduction in feral horses. Wildl Soc Bull 2 1 : 1 1 6- 1 2 1 . Estes RD, Estes RK, 1 979. The birth and survival of wildebeest calves. Z Tierpsychol 50:45-95 . Fairbanks LA, 1 996. Individual differences in maternal style. Causes and consequences for mothers and offspring. Adv Stud Behav 25:579-6 1 1 . Fairbanks LA, McGuire MT, 1987. Mother-infant relationships in vervet monkeys: Response to new adult males. lnt J Primatol 8:35 1 -366. 82 Green WCH, 1 992. The development of independence in bison: pre-weaning spatial relations between mothers and calves. Anim Behav 43:759-773. Grinnell 1, McComb K, 1 996. Maternal grouping as a defense against infanticide by males: evidence from field playback experiments on African lions. Behav Ecol 7 :55-59. Harrington FH, Mech LD, Fritts SH, 1 983. Pack size and wolf pup survival: their relationship under varying ecological conditions. Behav EcoI Sociobiol 1 3 : 1 9-26. Hauser MD, 1 988. Variation in maternal responsiveness in free-ranging vervet monkeys : a response to infant mortality risk? Am Nat 1 3 1 :573-587. Henneke DR, Potter GD, Kreider JL, Yeates BF, 1 983 . Relationship between condition score, physical measurement and body fat percentage. Equine Vet 1 1 5 :37 1 -372. Hinde RA, Atkinson S, 1 970. Assessing the roles of social partners in maintaining mutual proximity, as exemplified by mother-infant relations in rhesus monkeys. Anim Behav 1 8 : 1 69- 1 76. Hood, LC, 1 994. Infanticide among ringtailed lemurs (Lemur catta) at Berenty Reserve, Madagascar. Am J Primato1 33 :65-69. Hrdy SB, 1 977. Infanticide as a primate reproductive strategy. Am Sci, 65: 40-49. Hrdy SB, Hausfater G, 1 984. Comparative and evolutionary perspectives on infanticide: introduction and review. In: Infanticide: Comparative and Evolutionary Perspectives (Hausfater G, Hrdy SB eds). New York: Aldine; xiii-xxxv. lohnson RL, Southwick CH, 1 984. Structural diversity and mother-infant relations among rhesus monkeys in India and Nepal. Folia Primato1 43: 1 98-2 1 5. loubert E, 1 972. The social organization and associated behavior in the Hartmann zebra, Equus zebra hartmannae. Madoqua 1 : 1 7-56. Kappeler PM, 1 997. Determinants of primate social organisation: comparative evidence and new insights from Malagasy lemurs. BioI Rev 72: 1 1 1 - 1 5 1 . Kaseda Y, Khalil AM, Ogawa H, 1 995. Harem stability and reproductive success of Misaki feral mares. Equine Vet 1 27:368-372. Kiltie RA, 1 982. Intraspecific variation in the mammalian gestation period. J Mammal 63: 646-652. Kirkpatrick JP, Turner JWlr, 199 1 . Changes in herd stallion among feral horse bands and the absence of forced copulation and induced abortion. Behav EcoI Sociobiol 29: 2 1 7-2 19 . Kolter L, Zimmerman W, 1 988. Social behavior of przewalski horses (Equus p. przewalskii) in the Cologne Zoo and its consequences for management and housing. Appl Anim Behav Sci 2 1 : 1 1 7- 145. Leland L, Struhsaker TT, Butynski TM, 1984. Infanticide by adult males in three primate species of Kibale Forest, Uganda: A test of hypotheses. In: Infanticide: Comparative and Evolutionary Perspectives (Hausfater G, Hrdy SB eds). New York: Aldine; 1 5 1 - 1 72. Linklater WL, 1998. Social and spatial organisation of horses. PhD thesis, Massey University, New Zealand. 83 Linklater WL, Cameron EZ, Stafford KJ, Austin T, 1 998. Chemical immobilisation and temporary confinement of two Kaimanawa feral stallions. N Z Vet J 46: 1 17 - 1 1 8 Lycett JE, Henzl SP, Barrett S , 1 998. Maternal investment in mountain baboons and the hypothesis of reduced care. Behav EcoI Sociobiol 42:49-56. Maestripieri D, 1 993. Maternal anxiety in rhesus macaques (Macaca mulatta). n. Emotional bases of individual differences in mothering style. Ethology 95:32-42. Maestripieri D, 1 995. Assessment of danger to themselves and their infants by rhesus macaques (Macaca mulatta) mothers. J Comp Psychol 109:4 1 6-420. Maestripieri D, 1 998. Social and demographic influences on mothering style in pigtail macaques. Ethology 104:379-385. Mallory FF, Brooks RJ, 1 978. Infanticide and other reproductive strategies in the collared lemming, D icrostonyx groenlandicus. Nature 273 : 144- 146 . McLean IG, 1983 . Paternal behavior and killing of young in Arctic ground squirrels. Anim Behav 3 1 : 32-44. Monard A-M, Duncan P, Fritz H, Feh C, 1 997. Variations in the birth sex ratio and neonatal mortality in a natural herd of horses. Behav Ecol SociobioI 4 1 :243-249. Monard A-M, Duncan P, Boy V, 1 996. The proximate mechanisms of natal dispersal in female horses. Beh 1 33 : 1 095- 1 124. Oftedal OT, 1 985. Pregnancy and lactation. In: Bioenergetics of Wild Herbivores (Hudson RJ, White RG eds). Boca Raton, Florida: CRC Press; 2 1 5-238 . Packer C , Pusey AB, 1 983 . Adaptations of female lions to infanticide by incoming males. Am Nat 1 2 1 : 7 1 6-728. Packer C, Pusey AB, 1 984. Infanticide in carnivores. In: Infanticide: Comparative and Evolutionary Perspectives (Hausfater G, Hrdy SB eds). New Yark: Aldine; 3 1 - 42. Palombit RA, Seyfarth RM, Cheney DL, 1 997. The adaptive value of 'friendships' to female baboons: experimental and observational evidence. Anim Behav 54:599- 6 1 4. Parmigiani S , Palanza P, Mainardi D, Brain PF, 1 994. Infanticide and protection of young in house mice (Mus domesticus): female and male strategies. In: Infanticide and Parental Care (Parmigiani S, vom Saal FS eds) . Chur: Harwood; 34 1 -363. Penzhorn BL, 1 984. A long-term study of social organisation and behavior of Cape mountain zebra Equus zebra zebra. Z Tierpsycho1 64:97- 146. 84 Pereira ME, Weiss ML, 1 99 1 . Female mate choice, male migration, and the threat of infanticide in ringtailed lemurs. Behav Ecol Sociobiol 28: 1 4 1 - 1 52. Pusey AB, Packer C, 1 994. Infanticide in lions: consequences and counterstrategies. In: Infanticide and Parental Care (Parrnigiani S, vom Saal FS eds). Chur: Harwood; 277-299. RoIlinson DHL, Harker KW, Taylor JI, 1 956. Studies on the habits of zebu cattle. IV. Errors associated with recording technique. J Agric Sci Carnb 47: 1 - 5 . Rudman R, Keiper RR, 1 99 1 . The body condition of feral ponies on Assateague Island, Equine Vet J 453-456. Rutberg AT, 1 990. Intergroup transfer in Assateague pony mares. Anim Behav 40:945- 9 5 2 . Rutberg AT, Greenberg SA, 1 990. Dominance, aggression frequencies and mode of aggressive competition in feral pony mares. Anim Behav 40:322-33 l . Ryder OA, Massena R, 1 98 8 . A case of male infanticide in Equus przewalskii. Appl Anim Behav Sci 2 1 : 1 87- 1 90. SAS Institute Inc, 1 990. SAS/STAT User' s Guide, release 6. 4th ed. Cary, North Carolina: SAS Institute. Smith-Funk ED, Crowell-Davis SL, 1 992. Maternal behavior of draft mares (Equus caballus) with mule foals (Equus asinus X Equus caballus). Appl Anim Behav Sci 3 3:93- 1 1 9. Swenson JE, Sandegren F, Soderberg A, Bjarvell A, Franzen R, Wabakken P, 1 997. Infanticide caused by hunting male bears . Nature 3 86:450-45 1 . Trivers RL, 1 972. Parental investment and sexual selection. In: Sexual Selection and the Descent of Man (Carnpbell, B ed). Chicago: Aldine; 1 36- 1 79. Trivers RL, 1 974. Parent-offspring conflict. Am Zool 1 4: 2 1 9-26 l . TyIer SJ, 1972. The behavior and social organization of the New Forest ponies. Anim Behav Monogr 5 : 85- 1 96. Van Schaik C, Kappeler PM, 1 997. Infanticide risk and the evolution of male-female association in primates . Proc Roy Soc Lond B 264: 1 687 - 1 694. Whittemore CT, 1 980. Lactation of the Dairy Cow. New York: Longman. W right PC, 1 995. Demography and life history of free-ranging Propithecus diadema edwardsi in Ranomafana National Park, Madagascar. Int J Primatol 1 6 :835-854. CHAPTER 4 Social Environment SHARED OFFSPRING CARE BY A MOTHER AND HER DAUGHTER "BLOOD 'S THE WORD. NOTHING LIKE BLOOD, IN HOSSES, DA WGS, AND MEN" 85 Thackeray, 1847-1848, 'Vanity Fair'. 86 Photo caption: Toronto, with Banff (066). Banff is mother of Saskatchewan (Sassy, 021), and both mares shared suckling and care. Photo by Elissa Cameron 87 ABSTRACT Non-offspring nursing is characteristic of species that have large litter sizes (polytocous) and live in small stable kin groups. In monotocous species this extreme form of communal parenting is associated with stealing and exclusive adoption. Wild horses (Equus caballus) are monotocous and live in small, stable non-kin groups. We report a mother and daughter who raised a foal together, sharing and coordinating all offspring care. The foal received the same total input as foals reared singly but each mare provided less input. An examination of the communal suckling that occurs in such unusual circumstances helps to explain its evolution. Our results are similar to other reported cases of cooperative communal suckling in monotocous species, which seems more common than previously thought. In all cases the cooperating individuals were close relatives, indicating that high degrees of relatedness are a pre-requisite for the occurrence of communal suckling in monotocous species, and implicating kin selection in the evolution of cooperative communal care. C hapter reference: Cameron, E. Z., Linklater, W. L., Stafford, K. J. & Minot, E. O. Shared suckling and offspring care by mother and daughter feral horse mares. Submitted to Equine Veterinary Journal. INTRODUCTION Amongst mammals non-offspring nursing is the most extreme form of communal parenting. This is because lactation is the most energetically costly part of parental investment (Clutton-Brock, 1 99 1 ; Packer, Lewis & Pusey, 1 992) . Non-offspring nursing is most common in species characterised by large litters and small kin groups (Packer et al. , 1 99 2 ; ego lions Panthera leo Pusey & Packer, 1 994) . Although non-offspring nursing has also been reported in monotocous species (eg. water buffalo Bubalus bubalus, Murphey et al. , 1 995; African elephant Loxodonta africana, Dublin, 1 983; Lee 1 987; Indian elephant Elaphus maxim us, MacKay, 1 973; Rapaport & Haight, 1 987; fallow deer Cervus dama, San Jose & Braza, 1 993; bighorn sheep Ovis canadensis, Hass, 1 990) it i s almost always associated with reproductive errors (Riedman, 1 982) such as milk theft or exclusive adoption (Packer et ai. , 1 992) . Simultaneous non-offspring nursing in monotocous species has been reported in African elephants (Lee, 1 9 87), in some bat species that live in large colonies (eg. Mexican free-tailed bat Tadarida brasiliensis, McCracken, 1 984; pipistrelles Pipistrellus pipistrellus, Eales, Bullock & Slater, 1 988) and in captive Indian elephants (Rapaport & Haight, 1 987). Recent research suggests that nutritive non-offspring nursing in African elephants is rarer than previously thought as 8 8 most reported instances were probably non-lactating juveniles allowing infants to suckle (Lee & Moss, 1 98 6 ; Lee, 1 987; Lee, 1 989) . Horses are a monotocous species (Platt, 1 978) and non-offspring nursing is rare both in captive (Crowell-Davis, 1 985) and feral populations (Tyler, 1972; Packer et al. , 1 992). Mares are typically intolerant of non-offspring foals and of other mares that approach their own young foals . Foals, however, may attempt to suck from mares that are not their mother (Crowell-Davis, 1 985), from sub-adult females and even from maturing or adult males (Tyler, 1 972), mares are typically intolerant of non-offspring foals and of allowing other mares to approach their own young foals (Tyler, 1 972; Feist & McCullough, 1 976; Berger, 1 986) . The occasional instances of non-offspring nursing are characterised by short suckles and involve mistaken identity, stealing or attempted stealing (Tyler, 1 972; Crowell-Davis, 1 98 5 ; Berger, 1 986). Adoption or swapping of offspring are rarer still and only observed in domestic horses where the mothers are confined and parturition is highly synchronised (Tyler, 1 972; Crowell-Davis & Houpt, 1 986; Huntingdon & Cleland, 1 992), or with human intervention (Tyler, 1 972). All these instances are regarded as cases of reproductive error (Riedman, 1 982). Shared care and nursing of an offspring by mares has not before been reported in equids. We measured the behaviour of a mother and daughter Kaimanawa horse who raised a single foal between them, and compared their behaviour to mares who raised their foals singly. We also recorded instances where shared suckling could have occurred, such as orphaning and foal loss, to determine the circumstances in which shared care occurs. METHODS We have been studying the maternal behaviour of wild horses in the Kaimanawa ranges, New Zealand since 1 994, and have recorded over 3000 suckle bouts involving 1 1 3 foals . Of these only one foal (F) has been observed to suck from more than one mare. F consistently nursed from two mares, who were themselves mother (M, born 1 986, estimated from tooth wear) and daughter (D, born 1 993). It is unusual in feral horse populations for mothers and daughters to live together because daughters disperse around sexual maturity (Berger 1 98 6 ; Rutberg & Keiper, 1 993; Monard, Duncan & Boy, 1 996) . In the situation described M and D joined their current band in late 1 994 when D was a yearling still sucking from M. Consequently the stallion in the new band is not D's father, and M and D have remained in this band together. Both M and D were judged pregnant from assays of faecal estrone sulphate concentration in samples taken between March and September 1 996 (Henderson et al. 1 997). However, parturition by M and D was not seen and they were observed with only one foal between them when it was around 3 days old. 89 To determine if the behaviour of M, D and F was different from those of normal mare-foal dyads, each mare was paired with two control mares. The mares were paired on the basis of their age, parity and the sex of their offspring. Band size was also similar for each of the six mares. All six mares and their foals were sampled between 6 and 1 1 times from birth to 100 days of age in focal samples of 30 to 220 minutes. The total observation time is shown in Table 1 . Table 1 . Details of focal mares and foals. Symbol ID. Name Borna Band sizeS Ex£erience d .o .b . Obs D 02 1 S assy 1 993 8 Primiparous 2 . 1 1 .96 1 190 D1 009 Libby 1 993 5 Primiparous 6 . 1 1 .96 865 D2 1 69 Celeste 1 993 3 Primiparous 1 . 10 .96 1 060 M 066 Banff 1 986 8 Last foal 1 994 2 . 1 1 .96 1 1 90 M1 1 27 Ulysses 1 986 7 Last foal 1 996 1 8 . 1 0.96 570 M2 1 00 Darcy 1 987 7 Last foal 1 993 22 . 1 1 .96 7 10 a. Date of birth estimated by tooth wear patterns during a muster in 1 994 except for D2 who was known to be a yearling in 1 994. b. Number of adult females in group. c. Date of birth accurate to ± 2 days. d. Total time that behaviour was sampled (minutes) from birth to 1 00 days of age. During a sample we recorded every suck attempt; those lasting less than 5 seconds were denoted unsuccessful as no milk is likely to have been transferred in the first 5 seconds (Whittemore, 1 980). The length of successful sucks and the interval between sucks was recorded to the nearest second. Small breaks in nipple contact during a bout were subtracted from measures of suckle bout duration. We used average duration (excluding breaks in nipple contact within a bout) and average inter-suck time to calculate total time spent suckling (Becker & Ginsberg, 1990) . Bursts of sucking within a bout separated by breaks in nipple contact were termed episodes (termed bursts by Carson & Wood-Gush, 1983), and the number of episodes per bout was recorded. We also recorded whether or not a foal bunted (violently pushed nose into udder) during an episode and whether the mare of the foal terminated each episode. Breaks in nipple contact and bunting may indicate loss of milk flow or hunger (Lent, 1 97 1 ; Gomendio, 1 989), and the terminator of an episode may indicate levels of conflict between mare and foal and whether the foal fed to satiation (Byers & Bekoff, 1 990). We recorded every approach or leave event between mare and foal across a 2 body length boundary around each individual, and the instigator of these movements. These were used to calculate an index of the effort each put into maintaining contact (difference between % approaches and % leaves due to foal; Hinde & Atkinson, 1 970) . The index ranges between - 1 00 and + 1 00, with lower scores indicating more effort by the mother. - - - - - - - - - - - - - - - - - - - - ---------- 90 At four minute intervals we recorded instantaneous samples of the distance between mare and foal, and whether or not the mare and foal were nearest neighbours. We recorded the behaviour of the mare and foal, and for analysis of mare-foal distance and nearest neighbours we used only samples where at least one of the mare and foal were active (ie not standing or lying). Throughout the study we have also recorded orphaning of foals and the result of the orphaning, and foal loss by mares to determine if shared suckling occurred. RESULTS F sucked M and D alternately; 83% of suckles were followed by a suck from the other mare (X2 = 1 0. 60, p<0.0 1 ) . The total time F spent sucking per day (breaks in nipple contact between episodes of sucking within a bout were excluded), from M or D was less than the time control foals sucked from their mothers. Consequently, M and D each suckled F for less than half the time other mares suckled their foals, well outside the spread of times for all mares with female foals in 1 996 (M = 1 9.36 mins/day, D = 1 3 .38 mins/day; range for 20 other mares with female foals in 1 996 = 39.73- 1 46.09mins/day) . For F, each sucking bout contained fewer breaks in nipple contact (F, 48% 1 episode only, controls, 25% 1 episode only, X2=6.58, 1 df, p<0.02) , fewer sucking episodes were associated with bunts (X2=6. 1 5 , 1 df, p<0.02, figure l a), the proportion of unsuccessful sucks was not significantly different but tended to be less (X2=0.624, 1 df, ns, figure I b) and fewer sucking episodes were ended by the mother (X2=7.54, 1 df, p 1 5 body lengths, Fig 2d) from their foals after the first 20 days. However, the mean distance between mare and foal was similar for all ages during the first 50 days, with older mares being further from their foals thereafter (Fig 2b) . There was no significant difference with mare age in any suckling behaviour except the proportion of suckle attempts that were unsuccessful; older mothers had the least and mid-aged mothers the most unsuccessful suckles (Fig 2c) . Therefore, older mares showed the least conflict with their foals over the milk supply. There was no difference in rate of interaction between mares and their foals in relation to mare age although old mares tended to interact less (mean ± standard error: Y 0.33 ± 0.6 n=3 1 , M 0.3 1 ± 0.44 n=26, 0 0.24 ± 0.05 n=22, ANOV A F2.76=0.77 NS) . 1 04 1 .0 0.8 '" Q) ..... ro E 0.6 '- 0 c: .g 0.4 El P-o ..... c- 0.2 0.0 n=52 s ingle- stallion n=20 n==1 2 multi­ s tallion Band type Mare age o young � mid-aged . old Figure 1 . B and type in which mares raised their foals in relation to mare age. There was no significant difference in weaning age between age categories (ANOVA F2.3s=0.85, Fig 3 a). However, both young and mid aged mares weaned foals younger if they had a foal in the subsequent year than if they did not foal (young, tI 5==3 .8, P <0.0 1 ; mid, t1 3=2 .43, P <0.05) whereas there was no significant difference in foal weaning age in relation to the mother's future reproduction in old mares (t9== 1 .47, NS; Fig 3b) . Old mothers generally weaned their foals before one year of age regardless of whether they were pregnant, whereas younger mothers continued to suckle their yearlings if they did not foal. Mare age had no significant effect on differences in mare weight change either during gestation (mid gestation condition minus pre-birth condition, ANOVA F2.57== 1 . 1 2 NS) or during from birth to after peak lactation (pre-birth condition minus condition during 3rd month post-partum (ANOVA FZ.74= 1 .37 NS). Of all the mares that produced a foal in one year only a proportion also foaled in the subsequent year. We found no significant difference in this proportion between young, mid and old mares, though there was a tendency for more mares to foal as they aged (X2= 5 . 05 , 2df, 0.05< P <0. 1 , Fig 4). Mare age was a significant predictor of whether a mare foaled in the year after successfully raising a foal (Logistic regression X2= 6.45, 1 df, P <0.0 1 ) . - - - -- -- - -- - -- - -- - - - 1 05 a 0.8 >< <) 1 "Cl 0.6 • c ,'" - - - C) , ,f ' / / t , u § c 0.4 C) E .c;; E U 0.2 ' I ro r E 0 r u 0 b 1 0 ca 8 ,£ c Mare age ...... /"', u Vl 6 ---e- young .... ..c t<:I -;::: 0Il �mid-aged - ::: '" '" - - . - old w - c >. 4 <'l "Cl I to o I .- ..0 "Cl '-' c:: 2 t<:I <:.l E 0 c 0.3 2 Vl Vl '" 0.25 u u ::l Vl ::: 0.2 ::l Vl a E 0 . 1 5 B '" � 0 . 1 i u ::l Vl is 0.05 "€ 0 0. 0 0 ... 0... 0-20 2 1 -50 5 1 - 1 10 1 10-200 20CH- Foal age in days Figure 2. Mare effort with increasing age up to weaning in tenus of a) maintaining contact with her foals (% approaches-% leaves by foal, Hinde and Atkinson 1 970), b) the mean distance between mare and foal, and c) the proportion of suckle attempts that were unsuccessful. Values presented are means ± standard error 1 06 n= 1 7 1 5 9 7 6 3 1 0 9 6 5 00 � '" ;;... 400 � S Mare age Oi) :::: 300 0 young :::: � f2l midaged Il) $ • old 'i;j 200 Il) Oi) <: 1 00 0 All bals No new hal New fat! Figure 3 . Weaning age of foals by the age of its mother for all foals, foals whose mothers did not foal in the subsequent year, and foals whose mothers foaled in the subsequent year. Values are means ± standard error. n= 80 76 69 3 1 32 26 80 Oi) :E 60 Mare age <:tl <2 o young Vl Q) 1-< � mid-aged � 40 E E • old Il) 8 C) 20 Cl.. 0 In any year Afterraising a fall Figure 4. Reproductive success of mares of different ages in any year, regardless of reproduction in the previous year, and in the year following successful foal rearing. 1 07 DISCUSSION Older females are more successful at raising offspring because they have a higher probability of giving birth in any year and lower death rates of neonates (eg. Ozoga and Verme 1 986; Festa-Bianchet 1 988a). Our results show that Kaimanawa mares were consistent with this pattern. Old Kaimanawa mares show greater effort in the first 20 days after birth, during which time most foal deaths occur, but less effort thereafter. Other studies have found similar results (eg. bison, Green 1 986, 1 990; vervet monkeys, Fairbanks 1 996) . In addition old mares, with the lowest residual reproductive value show the least conflict with their foals over suckling, whereas mid aged mares show the most conflict. These results are consistent with greater experience leading to better targeted investment; older mothers appear to be better mothers due to their past experience. Older mares appear to be investing no more and possibly less, in total . However, the nutritional demands of lactation are by far the greatest cost to mothers, and we did not assess these here. Nonetheless , there was some indication that older mares were more tolerant of foal suckling, consistent with their decreasing reproductive value. These results are consistent with the influence of foal loss on subsequent reproductive effort in Kaimanawa horses; mares that lost a foal increased their effort during the first 20 days of foal life (Carneron 1 998 [chapter 3]) . Maternal investment is defined in terms of costs to future reproduction. Therefore, we can distinguish between the increased maternal investment predicted by RRV models and the lack of increased investment that would be predicted from TRE models of increasing maternal experience. We found no difference in costs in terms of weight change between the three age categories. Old mares, however, were more likely to foal in the year following the successful rearing of a foal. In contrast, only about half of the young mares foaled after successfully rearing a foal in the previous year. This result should be treated with caution. However, due to the inaccuracy of aging techniques (Richardson et al. 1 995) means that the range of ages in the old category was greater than in the other categories, and the very old mares may have suffered some reproductive costs . In domestic horses a decline in reproductive success occurs after around 1 5 years of age, though it is not clear whether this is an effect of age or multiparity (Ginther 1 992). S imilar declines have been noted in some feral populations (eg. Berger 1 986, Garrott et al . 1 99 1 ) , but not in others (eg. Duncan 1992). Nevertheless, the hypothesis of increased maternal effort with declining residual reproductive value predicts that maternal investment increases as females age, not just in very old mothers (Clutton-Brock 1 99 1 ), although most strong evidence of increased maternal investment with age occurs in mothers who do not survive to reproduce again (eg. Clutton-Brock 1984, Green 1 990). Therefore, our results support predictions based on increasing experience rather than declining residual reproductive value. This is consistent with other studies that have 1 08 shown that reproduction costs more to young mothers and less to old mothers (eg. Green 1 986, Festa-Bianchet 1 988b, Ktinkele and Kenagy 1 997). Older mothers are more successful mothers because they target their investment to the periods of high foal death and invest less thereafter. They wean all foals around the same age regardless of their pregnancy status, and suffer lower reproductive costs. They do not appear to invest more, but are 'older and wiser'. ACKNOWLEDGEMENTS This study was funded by a Department of Conservation contract to Massey University. We thank the New Zealand Army for granting permission to work in the Army Training Area, and for providing some logistic support. We particularly thank Keith Henderson, AgResearch Invermay for analysis of pregnancy and Clare Veltman for her input. Our experimental procedures were approved by the Massey University Animal Ethics Committee. REFERENCES Altmann J ( 1 974) Observational study of behaviour: sampling methods. Behaviour 49:227-267 Becker CD, Ginsberg J ( 1 990) Mother-infant behaviour of wild Grevy' s zebra: adaptations for survival in semi-desert East Africa. Anim Behav 40: 1 1 1 1 - 1 1 1 8 Berger J ( 1 979) Weaning conflict in desert and mountain bighorn sheep (Ovis canadensis) : an ecological interpretation. Z Tierpsycho1 50: 1 88-200 Berger J 1 986 Wild Horses of the Great Basin. Chicago University Press, Chicago. Boy V, Duncan P ( 1980) Time budgets of Camargue horses 1 . Developmental changes in the time-budgets of foals . Behaviour 7 1 : 1 87 -202 Byers JA, Bekoff M ( 1 992)Inference in social evolution theory: a case study. In: Bekoff M, Jarnieson D (eds) Interpretation and Explanation in the Study of Animal Behavior. Westview Press, Boulder. pp 84-97 Cameron EZ (in press) Is suckling behaviour a reliable predictor of milk intake? A review. Anim Behav Cameron EZ ( 1 998) Maternal investment in Kaimanawa horses. PhD thesis, Massey University, New Zealand. Cameron EZ, Stafford KJ, Linklater WL, Veltman Cl (in press) Suckling behaviour does not measure milk intake in horses . Anim Behav Carson K, Wood-Gush DGM ( 1 983) Behaviour of thoroughbred foals during nursing. Equine vet J 1 5 :257-262 Clutton-Brock TH ( 1984) Reproductive effort and terminal investment in iteroparous animals . Am Nat 1 23 : 2 1 2-229 - - -- �-- --- -- - - - - - 1 09 Clutton-Brock TH ( 1 99 1 ) The Evolution of Parental Care. Princeton University Press, New Jersey Crowell-Davis SL ( 1986a) Spatial relations between mares and foals of the Welsh pony, Equus caballus. Anim Behav 34: 1 007- 10 15 Dolhinow P, McKenna 11, Yonder Haar Laws J ( 1 979) Rank and reproduction among female langur monkeys: Aging and improvement (they're not just getting older, they' re getting better). Aggr Behav 5 : 1 9-30 Duncan P ( 1 992) Horses and Grasses. Springer-Verlag, New York Duncan P, Harvey PH, Wells SM ( 1984) On lactation and associated behaviour in a natural herd of horses. Anim Behav 32:255-263 Fairbanks LA ( 1 988) Mother-infant behavior in vervet monkeys. Response to failure of last pregnancy. Behav EcoI Sociobiol 23 : 1 57- 1 65 Fairbanks LA ( 1 996) Individual differences in maternal style. Causes and consequences for mothers and offspring. Adv Stud Behav 25:579-6 1 1 Festa-Bianchet M ( 1988a) Age-specific reproduction of bighorn ewes in Alberta, Canada. J Mamm 69: 1 57- 160 Festa-Bianchet M ( 1 988b) Nursing behaviour of bighorn sheep: correlates of ewe age, parasitism, lamb age, birthdate and sex. Anim Behav 36: 1 445- 1454 Garrott RA, Eagle TC. Plotka ED ( 1 99 1 ) Age-specific reproduction in feral horses. Can 1 ZooI 69:738-743 Green WCH ( 1 986) Age-related differences in nursing behavior among American bison cows (Bison bison). J Mamm 67:739-74 1 Green WCH ( 1 990) Reproductive effort and associated costs in bison (Bison bison) : do older mothers try harder? Behav Ec01 1 : 148- 1 60 Green WCH ( 1 993) Social effects of maternal age and experience in bison: pre- and post­ weaning contact maintenance with daughters. Ethology 93: 146- 1 60 Hinde RA, Atkinson S ( 1970) Assessing the roles of social partners in maintaining mutual proximity, as exemplified by mother-infant relations in rhesus monkeys. Anim Behav 1 8 : 1 69- 1 76 Ktinkele J, Kenagy G1 ( 1 997) Inefficiency of lactation in primiparous rats: the costs of first reproduction. Physiol Zoo 1 70:57 1 -577 Linklater WL ( 1998) Social and spatial organisation of horses. PhD thesis, Massey University, New Zealand. Meikle DB, Drickamer LC, Vessey SH, Arthur RD, Rosenthal TL ( 1 996) Dominance rank and parental investment in swine (Sus scrofa domesticus). Ethology 1 02 :969-978 Ozoga JJ, Verme LJ ( 1986) Relation of maternal age to fawn rearing success in white­ tailed deer. 1 Wildl Manage 50:480-486 Pianka ER, Parker WS ( 1 975) Age-specific reproductive tactics. Am Nat 109:453-464 1 10 Richardson JD, Cripps RJ, Lane JG ( 1 995) An evaluation of the accuracy of aging horses by their dentition: can a computer model be accurate? Vet Rec 1 37: 1 39- 140 Rollinson DHL, Harker KW, Taylor JI ( 1 956) Errors associated with recording technique. J Agric Sci 47: 1 -5 Rubenstein DI ( 1 982) Reproductive Value and Behavioral Strategies: Coming of Age in Monkeys and Horses. In: Perspectives in Ethology Vol. 1 5 . Bateson PPG, Klopfer PH (Eds), Plenum Press, New York SAS Institute Inc ( 1990) SAS/STAT User's Guide, Release 6. 4th ed. Cary, North Carolina: SAS Institute Smith-Funk ED, Crowell-Davis SL ( 1 992) Maternal behavior of draft mares (Equus caballus) with mule foals (Equus asinus X Equus caballus) . Appl Anim Behav Sci 33 :93- 1 1 9 Trivers RL ( 1 972) Parental investment and sexual selection. In: Campbell B (Ed.) Sexual Selection and the Descent of Man. Aldine, Chicago. pp 1 36- 1 79 Trivers RL ( 1974) Parent-offspring conflict Am Zool 14:2 1 9-261 Whittemore CT ( 1980) Lactation in the Dairy Cow. Longman, New York Williams GC ( 1966) Adaptation and Natural Selection. Princeton University Press, Princeton CHAPTER 6 Offspring Sex BIRTH SEX RATIOS RELATE TO MARE CONDITION AT CONCEPTION 1 1 1 HI FORMERLY THOUGHT THAT WHEN THE TENDENCY TO PRODUCE TWO SEXES IN EQUAL NUMBERS WAS ADVANTAGEOUS TO THE SPECIES IT WOULD FOLLOW FROM NATURAL SELECTION. " Darwin, 1871, 'The Descent of Man and Selection in Relation to Sex '. 1 1 2 Photo caption: Grommett (026) mates with Libby (009), with Libby 's foal Palace (96009) in the foreground. Photo by Elissa Cameron ABSTRACT Several hypotheses have been proposed to explain variation in birth sex ratios, based on the premise that variation is expected when the profitability of raising sons and daughters varies between individual parents. We test the Trivers-Willard hypothesis that mothers in better condition produce relatively more sons and that mothers in poorer condition produce relatively more daughters when male reproductive success is more variable. We examined birth sex ratios in relation to mare body condition at conception in horses. Horses meet the assumptions of the Trivers-Willard hypothesis better than many species on which it has been tested and in which sex ratio biases are not confounded by sexual size dimorphism such that one sex is more likely to be lost in utero by females in poor condition. Mares that had a female foal were in poorer condition at conception than those which had a male foal, and mares that had foals of different sexes in different years were in significantly poorer condition when they conceived their female foal. There was no relationship between offspring sex and mid-gestation condition and there was no difference in foaling rates in relation to body condition at conception. Consequently, sex ratio deviations are not explained by foetal loss in utero. Furthermore, differential foetal loss of the less viable sex cannot explain the greater proportion of males produced by mares in better condition. Therefore our results suggest that sex ratio modification occurs at conception in wild horses. Chapter reference: Cameron, E. Z., Linklater, W. L., Stafford, K. J. & Veitman, C. J. Birth sex ratios relate to mare condition at conception in Kaimanawa horses. submitted to Behavioral Ecology. INTRODUCTION Sex ratios at birth and hatching can vary (Clutton-Brock, 1 985; Clutton-Brock et aI. , 1 986; Clutton-Brock & Iason, 1 986; Frank, 1990; Hardy, 1 997). Adaptive theories predict variation in birth sex ratio when the profitability of raising sons and daughters varies between individual parents (Trivers & Willard, 1 973) . Large sex ratio variations were recently reported in Seychelles warbler eggs, indicating differential conception of male and female offspring in relation to offspring profitability (Komdeur et aI. , 1 997) . In mammals, although variations in birth sex ratios tend to be relatively small, some estimates have been based on inadequate evidence and there is a positive publishing bias (Festa-Bianchet, 1996), a few studies have shown significant variation (Clutton-Brock & Iason, 1986). The Trivers-Willard model suggests that where one sex has more variable reproductive success, such as males in polygynous species, mothers in good 1 1 3 1 14 condition will be advantaged by producing more of that sex, whereas mothers in poor condition would be advantaged by producing more of the reproductively stable sex (Trivers & Willard, 1 973) . The hypothesis was formulated for species with a litter size of one, and depends on three premises (Trivers & Willard, 1 973): a) that the condition of the young at the end of parental investment will tend to be correlated with the condition of the mother during parental investment. b) that these differences in condition tend to endure into adulthood. c) that adult males will be differentially helped in reproductive success by slight advantages in condition, such as through intense male-male competition in polygynous species. Wild horses meet these assumptions better than most species on which the hypothesis has been tested previously. Horse litter size is fixed at one (Platt, 1 978). Differential investment in male and female offspring has been reported (Berger, 1 986; Duncan et aI. , 1984), and there is a correlation between maternal rank and reproductive success in both sons (Feh, 1 990) and daughters (Duncan, 1 992), suggesting a correlation between female condition and offspring condition and reproductive success as adults. In some other species, one sex may be easier to invest in than the other. For example, where sons disperse and daughters are philopatric, it may be easier to influence the reproductive success of daughters for relatively little extra investment. This is not the case in horses, where both male and female offspring disperse, and so there can be no direct inheritance of social rank (Berger, 1986; Rutberg & Keiper, 1 993) . In addition, sexual size dimorphism at birth is minimal (Duncan, 1 992), so males are not markedly more costly to raise in utero due to accelerated growth rates alone, which has been a frequently cited reason why mothers in poor condition lose more male offspring in other species (Clutton-Brock, 1 99 1 ). Previous studies have examined offspring sex ratios in mammals in relation to indices of condition such as food availability or diet (eg. Monard et al ., 1997 ; Smith et al. , 1 996), maternal ranking (eg. Cassinello & Gomendio, 1 996; Clutton-Brock et al. , 1 984; Festa-Bianchet, 1 99 1 ), whether mothers were primiparous or mUltiparous (eg. Cassinello & Gornendio, 1 996), reproductive success in the previous year (eg. Green & Rothstein, 1 99 1 ; Rutberg, 1 986) , interval since last offspring (eg. Wiley & Clapham, 1993), or maternal body condition at slaughter when foetuses are mature enough to be sexed (eg. Reimars & Lenvik, 1 997; Wauters et aI. , 1 995). In Camargue horses, Monard et al. ( 1 997) found that sex ratios were female biased in years following a season of poor food availability, during which mares were in poorer body condition. However, they were unable to determine if differences were due to differential conception or to differential loss during gestation. Estimates of female body condition around (Kojola & Eloranta, 1 989) or before (Moses et aI. , 1 995) conception provide more appropriate estimates at the time when sex ratio adjustment can occur (Krackow, 1 995). We aim to determine if birth sex ratios vary according to the predictions of the Trivers-Willard hypothesis in relation to condition at conception both within and between individual mares. We examine differences in foaling rates of mares of different condition to determine if sex ratio variation is likely to be due to differential loss of offspring or differences in conception sex ratio. Furthermore, we investigate variation in sex ratio in relation to the previously used indices of whether mares are primiparous or multiparous, previous year' s foaling success and mid-gestation condition. METHODS Feral horses (Equus caballus), inhabit the Kaimanawa mountains and surrounding plateaus and valleys of central North Island, New Zealand. Since August 1 994 we have been studying a population of approximately 400 of these horses which inhabit the Moawhango river basin and surrounding plateau. Individual horses were reliably identifiable by either freeze brands on their rumps or by natural markings. Body condition scores were estimated by visual body fat distribution based on an 1 1 point scale from 0-5 with 0.5 gradations (Carroll & Huntingdon, 1 988; Huntingdon & Cleland, 1 992; Rudman & Keiper, 1 99 1 ) with the aid of 1O- 1 5x binoculars or a 1 5-60x telescope whenever horses were sighted, provided visibility was good. Mares with scores of 0 were very thin and with scores of 5 were obese. Interobserver reliability was high (EZC, WLL, r == 0.9 1 ) . In horses, body condition scores correlate with body fat percentage (r==0.8 1 , Henneke et al. , 1 983) . We calculated body condition at conception for Kaimanawa mares by backdating from the date of foaling (accurate to ± 5 days) by the average gestation length ± 1 standard deviation (336 ± 10 days; Kiltie, 1 982) and taking the mode of visual body condition scores during this period for mares with foals born in 1 995, 1 996 and 1 997, and at mid gestation ( 168 ± 10 days) for foals born in 1 995 and 1 996. All mares used in the analysis were scored at least twice during these 20 day periods, and mares were only scored once on any one day. Modal body condition scores were also calculated for mares in the month before birth, and for foals in their 1 2th month to determine if female condition during investment was reflected by foal condition near the end of the period of investment. Body condition scores at conception approximated a normal distribution with the modal scores of most mares being 2.5 and with a spread from 1 to 4. 33 mares foaled once, 30 foaled twice and 14 foaled three times between 1994 and 1 997. Mares were classified as primiparous if they had not foaled before and were known to be too young (� 2 years) to have foaled prior to the study (n==22) . Mares 1 1 5 1 1 6 were classified as multiparous only if they were known to have foaled previously (n= 1 03 ; unknown n= l O) . We also classified mares in relation to their previous year' s reproductive success (no foal or lost foal n=67, successfully raised foal to 6 months n=67, unknown n= 1 ) We recorded the sex of every foal born to a mare within the focal population in the foaling season (September to February) starting in 1 995 (n=42), 1 996 (n=55) and 1 997 (n=38) by sighting genitalia, which are visible in both sexes from birth. To determine if foaling rates were different between mares of different condition at conception we recorded the condition of all mares during the mean conception date ± 1 standard deviation ( 1 5 November to 1 6 January 1 994-95 and 1 995-96) and recorded which of these mares foaled in the subsequent season. The results of statistical tests presented are two-tailed. RESULTS There was a significant difference between the condition at conception of mares who had a female foal and mares who had a male foal (Mann-Whitney V-test, V=2794, N]=69, N2=66, p<0.05). Improved mare condition at conception was a significant predictor of offspring sex (logistic regression, df= l , X2=7.89, p- 80 '" Of) c: .v; 60 ';:a .... V> V .... 40 '" E � § u 20 "-v P- O 2 2.5 3 3 .5 Mare bcxiy mndition retOrebirth of lTevious son or drughter Figure 3 . Reproductive costs of sons (open bars) and daughters (filled bars) to mares of different condition. a) the proportion of mares that foal in the subsequent year in relation to her body condition before the birth of her previous son or daughter, and b) the proportion of mares that successfully raised a foal in the year after having a son or daughter in relation to her body condition before the birth of that son or daughter. foal in the subsequent year. Survival rates of foals also varied in relation to the sex of the previous year' s foal and the mother' s condition when she had the previous foal . Poor condition mares that had raised a daughter were less likely to successfully rear a foal in the subsequent year than were mares in better condition (d.f.=3 , X2=9.98, P <0.05, Fig 3b). Again the opposite trend was observed for sons (d.f.:::3 , X2:::6.67, P <0. 1 , Fig 3b), although all mares that scored 3.5 raised live foals. Therefore, sons and daughters have different reproductive costs for mares with different body conditions. 1 3 5 (c) Differential investment by individual mares When mares of the same condition but with different sexed foals were compared to themselves, mothers in better condition favoured their son, whereas mares in worse condition favoured their daughter. Using a combination of eight behavioural variables we found that mare condition was a significant predictor of the difference between a mare's behaviour towards her son and her daughter for the period of essential investment (birth to 1 10 days, MANOVA, Wilks' Lambda = 0.36, FS, 1 4 = 3 .05 , P <0.05) , and during the period when most foal mortality occurs (0-20 days, MANOVA, Wilks' Lambda = 0.34, FS, 1 2 == 2.89, P <0.05) . In addition, several variables were singly significant or approaching significance from birth to 1 lO days (Fig 4a) and from birth to 20 days of age (Fig 4b) . Mothers lost more condition pre-natally with a son if they were in good condition, but more with a daughter when they were in poor condition (ANOV A, F2, 1 1=3 . 8 1 , P <0.05, Fig Sa) . There were no significant difference in post-natal condition change with a son and a daughter in relation to the mare' s condition at foal birth (ANOVA F2,20==0.90, NS) . There was no difference in time between foaling and subsequent conception after sons and daughters, although the pattern was for a longer time after a daughter for poor condition mares and longer after sons for good condition mares (ANOVA, F2• 1 7== 1 .08, NS, Fig 5b). DISCUSSION We found no difference in maternal behaviour, when we looked for sex differences on a population level, supporting some previous studies (e.g . Crowell-Davis 1 985, Crowell­ Davis 1 986, Smith-Funk & Crowell-Davis 1 992), but contrary to others (Duncan et al. 1 984, Berger 1 986). However, both these latter studies base their conclusions on differences in time spent suckling which does not measure milk intake in horses and so inferences about milk intake based on time suckling are likely to be wrong (Cameron in press [chapter 1 ] , Cameron et al. in press [chapter 2]). It is not surprising that there is no difference between males and females on a population level as greater investment in all sons is predicted only for highly dimorphic species in which body size influences reproductive success (Maynard Smith 1 980). Convincing studies of male biased maternal investment on a population level have all been on highly dimorphic species (e.g. red deer Cervus elaphus Clutton-Brock et aL 1 984, 1986, bighorn sheep Ovis canadensis Berube et at. 1 996). Horses are not sexually dimorphic so population-based predictions do not apply. Indeed, if mothers in good condition favour sons, but that mothers in poor condition favour daughters, as the TWH predicts (Trivers & Willard 1973) than population differences will be obscured if mare condition is not controlled for. When mare body condition was taken into account we found that although 1 36 0.40 0 .24 0. 1 6 0.08 0.00 -0. 1 6 -0.24 0.4 -0.2 -0.3 % time close « 2bf) % time close « 2bl) % sucks ended by mother interaction % sucks Distance from unsuccessful mare to foal favours sons favours daughters favours sons favours daughters Figure 4. The difference between a mare's behaviour with her son and daughter in relation to her condition, a) from birth to 1 10 days, the period of essential maternal investment, and b) from birth to 20 days, the period during which most foal deaths occur. Poor condition mares n=7 (open bars), mid condition mares n= lO (hatched bars), good condition mares n=6 (filled bars). 0.8 c:: 0.6 .g 0.4 :;:; § 0.2 Cl .s Q 0 CD c:: " .c: -0.2 u -;; " -0.4 'i' v et -0.6 -0.8 I poor avemge 1 J good Mare body corditionscore Sons st rmre co D aughtelS oSt mcre c c: .s c.. \00 g 50 o Cl "0 § 0 Cil .s -;; -2 -50 § v " B - 1 00 '" >. '" Cl - \ 50 1 poor averag-: go