MATERNAL EFECTS IN COLUMBIAN GROUND SQUIRELS Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory commite. This thesis does not include proprietary or clasified information. ______________________ Amy L. Skibiel Certificate of approval: _________________________ ________________________ Robert S. Lishak F. Stephen Dobson, Chair Asociate Profesor Profesor Biological Sciences Biological Sciences _________________________ _________________________ Gary R. Hepp George T. Flowers Profesor Interim Dean Wildlife Sciences Graduate School MATERNAL EFECTS IN COLUMBIAN GROUND SQUIRELS Amy L. Skibiel A Thesis Submited to the Graduate Faculty of Auburn University in Partial Fulfilment of the Requirements for the Degre of Master of Science Auburn, Alabama May 10, 2007 ii MATERNAL EFECTS IN COLUMBIAN GROUND SQUIRELS Amy L. Skibiel Permision is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves al publication rights. ______________________________ Signature of Author ______________________________ Date of Graduation iv THESIS ABSTRACT MATERNAL EFECTS IN COLUMBIAN GROUND SQUIRELS Amy L. Skibiel Master of Science, May 10, 2007 (B.S., Juniata College, 2003) 74 Typed Pages Directed by F. Stephen Dobson The phenotype of an organism is the physical expresion of its genotype and is a result of both the genetic makeup of an organism and the environment it experiences. An individual?s phenotype can also be afected by the phenotype of its mother. A maternal efect occurs when the phenotype of the mother influences the phenotype of her offspring, independent of the offspring?s genotype or non-maternal aspects of its environment. In this study, maternal efects on pup growth and size in Columbian ground squirels were asesed. A cross-fostering study was used to determine the contributions of maternal efects and direct genetic efects to variation in pup traits. Maternal efects were responsible for a large proportion of the variation in ofspring growth rate and weaning weight, as indicated by the high asociations in these traits betwen unrelated litermates (growth rate: R 2 = 0.58; weight: R 2 = 0.64). Heritability estimates for both v offspring traits were zero, suggesting that maternal efects are a more important determinant of offspring phenotype than are genetic efects. Relationships betwen maternal phenotype and offspring growth rate and weaning weight and the influences of these traits on survival of pups over-winter were also examined. Liter size and maternal traits (body condition, timing of reproduction, structural size, and change in mas during reproduction) explained 69% of the variation in offspring weaning weight and 72% of the variation in offspring growth rate. Only maternal body condition and liter size had significant efects on ofspring weaning weight (condition: p = 0.34; liter size: p = -0.85) and growth (condition: p = 0.28; liter size: p = -0.88). Pup growth rate was the only variable with an efect on pup survival to yearling age. Influences of maternal investment in young did not appear to persist to yearling age, as suggested by the lack of asociation betwen maternal traits and size of offspring at emergence from hibernation the following spring (R 2 = 0.34). vi ACKNOWLEDGMENTS First and foremost I would like to thank my family for their suport and encouragement throughout the duration of my Master?s program. A special thanks to my advisor Dr. F. Stephen Dobson, for his guidance, statistical asistance, sugestions for improving the thesis, and for providing data from earlier years. I would also like to recognize my comite members, Dr. Gary R. Hep and Dr. Robert S. Lishak, my felow graduate students in the Dobson lab, and my friends, for their suport, words of wisdom, and helpful coments on the thesis. Apreciation is also extended to Dr. David Brousard who endured countles telephone cals regarding statistics and to Dr. Jan O. Murie for providing helpful coments on the manuscript and data from earlier years. A special thanks to the field asistants who have asisted in data colection over the course of the study. Jil Cordeiro also deserves recognition for data entry. The University of Calgary provided housing in the R. B. Miler field station during field research. Research funding was provided by a Graduate Research Award from the Graduate School at Auburn University and a U.S. National Science Foundation research grant (DEB- 0089473), awarded to F. S. Dobson. vii Journal format used: Ecology Computer software used: Microsoft Word 2002; Endnote 9.0; The SAS system for Windows 9.1 vii TABLE OF CONTENTS LIST OF TABLES.....................................................ix LIST OF FIGURES.....................................................x INTRODUCTION.....................................................1 CHAPTER ONE: MATERNAL EFECTS ON OFSPRING WEIGHT AND GROWTH RATE IN THE COLUMBIAN GROUND SQUIREL: A CROS- FOSTERING APROACH..............................................3 INTRODUCTION................................................4 MATERIALS AND METHODS.....................................6 RESULTS.....................................................11 DISCUSION..................................................14 CHAPTER TWO: EFECTS OF MATERNAL CHARACTERISTICS ON OFSPRING WEIGHT, GROTH RATE, AND SURVIVAL IN COLUMBIAN GROUND SQUIRELS.........................................................24 INTRODUCTION...............................................26 MATERIALS AND METHODS....................................29 RESULTS.....................................................35 DISCUSION..................................................39 CONCLUSIONS......................................................51 LITERATURE CITED.................................................55 ix LIST OF TABLES CHAPTER TWO, TABLE 1. AVERAGES OF OFSPRING TRAITS AND MATERNAL CHARACTERISTICS FOR EACH AGE CLAS..................44 CHAPTER TWO, TABLE 2. CORELATIONS OF VARIABLES USED IN PATH ANALYSES .........................................................46 x LIST OF FIGURES CHAPTER ONE, FIGURE 1. RELATIONSHIP BETWEN WEIGHTS OF UNRELATED LITERMATES AT WEANING............................20 CHAPTER ONE, FIGURE 2. PLOT OF ASOCIATION BETWEN FOSTER AND NON-FOSTERED PUP GROWTH RATES.................................21 CHAPTER ONE, FIGURE 3. REGRESION OF FOSTERED SIBLING WEIGHT ON NON-FOSTERED SIBLING WEIGHT....................................22 CHAPTER ONE, FIGURE 4. RELATIONSHIPS BETWEN ON-FOSTERED SIBLING ROWTH RATE AND FOSTERED SIBLING ROWTH RATE........23 CHAPTER TWO, FIGURE 1. PATH MODEL OF EFECTS OF MATERNAL CHARACTERISTICS ON OFSPRING TRAITS............................48 CHAPTER TWO, FIGURE 2. PATH MODEL WITH COEFICIENTS FOR THE INFLUENCE OF MATERNAL CHARACTERISTICS ON OFSPRING TRAITS...49 1 INTRODUCTION Maternal efects are any phenotypic variation that is due to the phenotype of the mother and exclusive of: 1) the genotype of the offspring (Bernardo 1996a, Roff 1998), 2) the non-maternal components of the ofspring's environment (Lacey 1998), and 3) the interaction of offspring genotype with its non-maternal environment (Bernardo 1996b). Until relatively recently, maternal efects were regarded as potentialy confounding factors in ecological and evolutionary studies of phenotypic variation and researchers atempted to design experiments that controlled for maternal efects (Bernardo 1996a). In the past several years literature pertaining to the topic of maternal efects has increased dramaticaly as researchers have become aware of the potential ecological and evolutionary ramifications of maternal efects themselves. For example, maternal care during early offspring development may increase survival probability of young through reproductive age, which could enhance both maternal and offspring fitnes. In this study we examined aspects of maternal efects in Columbian ground squirels. Specificaly, we sought to determine if maternal efects are responsible for observed variation in ofspring phenotypes (viz., growth rate and weaning weight). We investigated paterns of maternal investment on ofspring growth and size in the wild and searched for evidence of persistence of maternal influences on juvenile traits to yearling age. 2 The first objective was addresed by conducting a cross-fostering study, whereby neonates were transfered betwen pairs of liters. Thus, each liter consisted of a mother?s own pups and fostered pups. By examining relationships betwen unrelated litermates and betwen genetic siblings raised by diferent mothers, it was possible to disentangle contributions of both maternal efects and direct genetic efects (viz., heritability) to offspring traits. We predicted that if maternal efects had an influence on development, then unrelated litermates would be more closely related in growth rate and weaning weight than to genetic siblings that were raised in diferent liters. However, if heritability was more important than maternal efects in pup development, we expected genetic siblings raised in diferent liters to be more similar in traits than unrelated litermates. The second objective of the study was to describe specific efects of maternal traits on offspring phenotype and their persistence on offspring the folowing year. We first examined paterns of maternal investment in ofspring growth rate and weaning weight using a multivariate approach. We then searched for influences of maternal traits and offspring traits on survival of pups to yearling age and on pup weight at yearling age, as an indication of persistence of maternal influences on pup phenotypes. 3 CHAPTER ONE MATERNAL EFECTS ON OFSPRING WEIGHT AND GROWTH RATE IN THE COLUMBIAN GROUND SQUIREL: A CROS-FOSTERING APROACH Abstract. The external expresion of a trait is influenced by both the genetic composition of the individual as wel as the environment it experiences. Maternal efects are a specific source of this environmental variation and occur when the phenotype of the mother influences the phenotype of her offspring, independent of the ofspring's genotype. A potentialy important maternal efect in Columbian ground squirels is development of young prior to their first hibernation. The goal of this study was to determine the relative contributions of the genetic composition of offspring and maternal efects to pup weaning mas and growth rate in mas during the lactation period. A cross-fostering technique was employed to separate maternal efects from direct genetic efects on offspring phenotype. Maternal efects contributed significantly to variation in both weaning weight (R 2 = 0.64) and growth rate (R 2 = 0.58). The estimated heritability for weaning weight and growth rate was zero. This indicates that the environment provided by the mother has more of an influence on pup weight than direct genetic inheritance. 4 INTRODUCTION Natural selection acts on phenotypic variation (Stearns 1992), which is comprised of both genotypic and environmental factors (Bull 1987). Genotypic variance is due to diferences among individuals in their genetic composition. The environmental component of phenotypic variation is due to al non-genetic sources and arises when individuals experience diferent environments, such as nutritional or climatic diferences (Falconer and Mackay 1996). One particularly interesting source of environmental variation is maternal efects. Maternal efects are phenotypic influences of the mother on the phenotype of her offspring (Falconer and Mackay 1996, Bernardo 1996a, Mousseau and Fox 1998) and have been explored in a wide range of taxa, including plants (reviewed in Roach and Wulf 1987), insects (reviewed in Mousseau and Dingle 1991, Kyneb and Toft 2006), amphibians (Parichy and Kaplan 1992, Kaplan 1998), fish (Heath and Blouw 1998), birds (reviewed in Price 1998, Blums et al. 2002), reptiles (Kolbe and Janzen 2001) and mamals (Elis et al. 2000, McAdam et al. 2002). Maternal efects can be mediated through behaviors, such as nutritional provisioning of young pre-and post-partum (Mather and Jinks 1971), oviposition or nest site selection (Kolbe and Janzen 2001), preparation of nests and burrows, and protection against predators (Clutton-Brock 1991). Maternal influences also can occur physiologicaly, through transmision of pathogens, antibodies (Mather and Jinks 1971) and hormones (Mousseau and Dingle 1991, Clark and Galef 1995) from mother to ofspring. The result is modification of offspring traits, such as sex ratio (Trivers and Wilard 1973, Nager et al. 1999), liter size (Campbel and 5 Slade 1995, Dobson et al. 1999), propagule size (reviewed in Bernardo 1996b), and offspring quality (Elis et al. 2000). Ofspring quality impacts juvenile performance, such as growth and competitive ability, and survival (Bernardo 1996a). The potential for maternal influences on offspring phenotype should be especialy high during early ofspring development, which is often the case (Shaw and Byers 1998). In many mamalian species, young are altricial at birth and are dependent on parental provisioning, alowing ample opportunities for maternal efects to occur. In Columbian ground squirels, Spermophilus columbianus, young develop in a nest betwen birth and weaning and mothers provide the sole source of nutrition during this time. Thus, development of young, as reflected by growth rate and body mas, should be strongly influenced by maternal efects. However, variation in body mas and growth rate may also be atributable to diferences in the genetic composition of young. In this study, a cross-fostering technique was used to explore the relative contributions of direct genetic efects and maternal efects to ofspring weaning weights and growth rate in body mas in Columbian ground squirels. Cross-fostering is commonly applied to studies regarding evolution of clutch size in birds and can be readily applied to any species exhibiting parental care (Bernardo 1996a). Recent studies have succesfully used this method to detect maternal efects on juvenile growth rates, survival, and progeny size (McAdam et al. 2002, Pelayo and Clark 2003, Crespi and Lesig 2004). By using the cross-fostering technique, it is possible to discern direct efects of genes and indirect efects, due to the environment provided by the mother, on offspring phenotype (Roff 1998). 6 If weaning weights and growth rates are influenced by maternal efects, we predict that unrelated individuals within cross-fostered liters wil be similar in growth rate and weight at weaning and heritability of these traits should be low. That is, the proportion of the offspring phenotype atributable to the additive efects of genes should be low. However, if maternal efects are not influencing pup phenotype, heritability estimates should be large and unrelated litermates wil be diferent with respect to these traits. MATERIALS AND METHODS Study organism Columbian ground squirels are smal, iteroparous, hibernating rodents with a short active season, in which females have to mate, give birth, lactate, and acquire fat stores before the 8-9 month hibernation period (Murie and Boag 1984, Festa-Bianchet and King 1991, Dobson et al. 1992, Wilson and Ruff 1999). Females mate a few days after spring emergence from hibernation and produce one liter per year (Hare and Murie 1992). Life histories are extremely plastic as evidenced by early maturation in females, increased survival, larger liter size, and heavier individuals in populations supplemented with food (Dobson and Murie 1987). Population size is regulated, at least in part, by availability of food resources (Dobson 1995, Dobson and Oli 2001). 7 Study location Field studies of Columbian ground squirels were conducted during the summers from 1992-2006. This study includes data only from years 1993, 1994, 1999, 2001, and 2003; the years when pups were cross-fostered. This population inhabits a meadow (elevation of 1550m; 110?W 50?N) along the Sheep River in the Sheep River Provincial Park, Alberta, Canada. Trapping and experimental manipulation Ground squirels were captured at spring emergence (April-May) by placing live- traps (Tomahawk #201 collapsible chipmunk trap, 16 x 5 x 5 cm 3 ) baited with peanut butter near the opening of the burrow. Al individuals were marked with numbered fingerling ear tags and were given unique body markings using hair dye (Lady Clairol Hydrience #52 black pearl). Weight, using a Pesola scale, reproductive status, and zygomatic arch breadth were recorded for al captured individuals. During June and early July, juveniles were captured at emergence from natal burrows using the same techniques as described above. Time of liter emergence was estimated by adding 51 days to the mating date: 24 days from mating to parturition (Murie and Haris 1982) and 27 days from parturition to liter emergence (Murie 1992). Mating date was determined by examination of external morphology or by observations of pre- and post-copulatory behavior. For example, prior to mating the vulva becomes swollen and within 1 to 2 days following copulation the vulva is flacid (Murie and Haris 1982). Although most copulation occurred underground, pre- and post-copulatory behavior was observed 8 frequently. Males entering burrows with estrous females followed by female aggresion towards the male or cesation of male interest in a female (viz., snifing the anogenital region, kising) indicated that copulation had occured. Females that copulated were captured prior to parturition (May-June), weighed, and housed in the laboratory until birth (1-10 days). Females were kept in polycarbonate microvent rat cages (Alentown Caging Equipment Company; 267 x 483 x 203 m 3 ) on pine chip bedding and given newspaper for nest building material. To obscure vision from neighboring females and to simulate the burrow environment, cages were covered in black plastic bags and stored in a field lab maintained at ambient temperature. Squirels were fed a high protein horse fed (minimum crude protein 13%, oats, barley, wheat, and compresed vegetable material in a molases mix) ad libitum and letuce and apple twice daily. Females were checked for pups 3-4 times per day from 0530-2230 hrs. Females and neonates were weighed at least 4 hours after birth (to ensure that parturition was complete) using a Pesola scale (mothers) or a Metler balance (neonates). Neonates were marked by removing a smal amount of tisue from the outer left or right toe bud or from the end of the tail. This alowed identification of pups when they emerged from natal burrows. Liters were then paired and one to thre neonates from each liter were exchanged. The same number of pups was cross-fostered betwen liters and each mother retained at least one of her own pups. Mothers were released on the meadow by opening the cage containing mother and pups. Mothers entered the nest burrow, usualy within 20 minutes. Pups were then placed in the nest burrow opening and the opening was observed until the mother retrieved her pups. Nest burrows are inconspicuous 9 burrows that females create before giving birth. Mothers give birth to and nurse pups in the nest burrow. Nest burrows were discovered by observing females stocking them with nesting material and were flagged prior to placing females in the lab. Statistical analysis Diferences in pup traits among years were examined by conducting analysis of variance (ANOVA) tests. Pup traits included weaning weight, birth weight, and growth rate. Pup data were averaged within liters to retain independence of data points. Weaning weights, birth weights, and growth rates did not difer significantly among years (respectively, R 2 = 0.10, F [ 4, 46 ] = 1.28, P = 0.29; R 2 = 0.13, F [3, 37 ] = 1.8, P = 0.16; R 2 = 0.06, F [ 3, 36 ] = 0.79, P = 0.51). Therefore, offspring data were pooled among years. Analyses were run using only 2001 (the year with the greatest sample size; n = 30 cross- fostered liters) and al years combined (n = 51 cros-fostered liters). Results did not difer; thus, results from al years combined are presented. Pooling data among years invokes multiple records for some females (n = 9 females), violating the asumption of data independence. However, records for females were considered independent because studies show that reproduction in one year does not influence liter size (Risch et al. 1995), liter mas (Dobson et al. 1999, Broussard et al. 2005a) or female survival to the following year (Murie and Dobson 1987). In addition, average offspring mas for females that were older than 1 year, did not significantly afect average offspring mas the following year (Spearman rank correlation; r = 0.17, n = 92, P = 0.10). 10 Regresions were employed to analyze relationships betwen traits of foster pups (e.g. pups that were fostered into that mother's liter; hereafter refered to as foster pups) and non-fostered pups (e.g. pups that remained with the birth mother; hereafter refered to as non-fostered pups). Only females that had at least one of her own and at least one fostered pup survive to weaning were used in analyses. When more than one foster or own pup survived to weaning, traits were averaged within liters. Growth rate indicates weight gain in grams per day during the lactation period and was calculated by subtracting mas at birth from mas at weaning and dividing by length (in days) of the lactation period. Duration of lactation is equal to the date of liter emergence minus liter birth date. Liter emergence date gives an approximate date for the termination of the lactation period (Anderson et al. 1976, Murie and Dobson 1987, Michener 1989, Rieger 1996). Since liter size has been shown to be correlated with pup weights (Michener 1989, Dobson et al. 1999), regresions were run both including and excluding liter size at weaning as a covariate. Estimates of heritability for weaning weight and growth rate were calculated from relationships betwen genetic siblings. Weaning weights and growth rates were standardized by taking residuals obtained from analysis of variance (ANOVA) tests of fostered sibling and non-fostered sibling weaning weights and growth rates using size of their respective liter at weaning as the grouping factor. When more than one fostered sibling or non-fostered sibling survived to weaning, weaning weights and growth rates were averaged within liters. Regresions used standardized weaning weights and growth rates of fostered siblings (e.g. siblings that were fostered to another liter; hereafter 11 refered to as fostered sibling) on non-fostered siblings (e.g. siblings that remained with the natal mother; hereafter refered to as non-fostered sibling). The slope of the line generated by the regresion of related individuals provided an estimate of the amount of additive genetic variance of a trait as a proportion of the total phenotypic variance (e.g. heritability; Falconer and Mackay 1996). In Columbian ground squirels, siblings share either 25% or 50% of their genes because multiple paternities within a liter are possible (Murie 1995); therefore, the slope of the regresion line only estimates 0.25 (for half sibs) or 0.50 (for full sibs) of the heritability of the trait . Slopes of the lines and standard erors can be corected by multiplying by 2 and 4. This gives a range within which heritability of offspring weight should fal. Al analyses were performed using SAS statistical software for Windows (SAS, 2002). Significance level for al tests was ! = 0.05. Data were tested for normality by performing Shapiro-Wilks tests. Al variables were normaly distributed. Linearity and homoscedasticity were examined graphicaly. Durbin-Watson tests were used to detect autocorrelation of residuals. RESULTS Liter size at weaning and non-fostered pup weaning weights explained a significant amount of variation in foster pup weaning weights (R 2 = 0.65, d.f. = 2,48, P < 0.0001; means ? 1 SE: liter size: 3.22 ? 0.11; non-fostered pup weights: 104.47 g ? 2.34; foster pup weights: 102.88 g ? 2.35). Non-fostered pup weaning weight alone explained only slightly les variation in foster pup weaning weight and the asociation 12 remained highly significant (Fig. 1; R 2 = 0.64, d.f. = 1,49, P < 0.0001). Non-fostered pup growth rate was 3.57 g/day ? 0.09 and foster pup growth rate was 3.51 g/day ? 0.09. Non-fostered pup growth rate explained 58% of the variation in foster pup growth rate (Fig. 2; d.f. = 1,35, P<0.0001). Inclusion of liter size at weaning into the regresion model as a covariate did not result in explanation of more of the variation in foster pup growth rate (R 2 = 0.58, d.f. = 2,34, P < 0.0001). The relationships betwen growth rate and weaning weight of non-fostered and foster pups within liters could have been due to an initial relationship betwen foster and non-fostered pup birth weights that persisted to weaning rather than due to maternal efects during postnatal development. However, non-fostered pup birth weight could only explain a smal and insignificant amount of variation in foster pup birth weight, both when liter size at weaning was included as a covariate and when liter size was excluded (respectively; R 2 = 0.06, d.f. = 2,35, P = 0.32; R 2 = 0.01, d.f. = 1,36, P = 0.63). This suggests that the relationship betwen traits of unrelated litermates arose during the lactation period. For both the regresion of foster pup weaning weight on non-fostered pup weaning weight and foster pup growth rate on non-fostered pup growth rate, the slopes of the estimated regresion lines (weight: b = 0.80 ? SE 0.09; growth rate: b = 0.76 ? SE 0.11) were significantly diferent from a line of slope of 1.0, which is predicted if foster pups and non-fostered pups are identical in weaning weight and growth rate (weight: F [1,49] = 5.27, P = 0.03; growth rate: F [1,35] = 4.71, P = 0.04). This indicates some 13 diferentiation betwen foster pup and non-fostered pup weaning weights and growth rates. The diferentiation betwen non-fostered pups and foster pups could be due to initial diferences in birth weights that persisted to weaning or to mothers diferentialy investing in pups she gave birth to. At birth, foster pups weighed an average of 12.24 g ? 0.17 and non-fostered pups weighed 11.98 g ? 0.14. Birth weights of foster pups were not significantly diferent from non-fostered pups (paired t-test: t = 1.38, d.f. = 37, P = 0.18). Weaning weights for both foster pups and non-fostered pups were not correlated with their birth weights (foster pups: r = 0.12, n = 39, P = 0.29; non-fostered pups: r = 0.27, n = 40, P = 0.09). If mothers were diferentialy investing in their own young over the fostered young, we would expect deviations of data points from the line of slope 1.0 to be non- normaly distributed or for non-fostered young to have increased survival to yearling age. Deviations were estimated by taking the distance of a perpendicular line from each data point to a line with a slope of 1.0 and an intercept of 0. T-tests were then performed to test if the deviations were significantly diferent from 0. Deviations of the data points from a line with a slope of 1.0 did not difer significantly from 0 for both weaning weight (t = -1.08, d.f. = 50, P = 0.29) and growth rate (t = -0.64, d.f. = 36, P = 0.53), indicating that females do not diferentiate betwen her own pups and the foster pups. A mixed model logistic regresion was used to test survival diferences in non-fostered and foster pups. Liter number, which was arbitrarily asigned to each liter, was included in the model as a random variable to control for data on multiple pups within the same liter. 14 Survival of pups to yearling age was not related to whether they were fostered or not (F [1,162] = 0.06, P = 0.80). Non-fostered siblings weighed an average of 103.74 g ? 2.40 at weaning and fostered siblings weighed 107.42 g ? 3.25. For siblings (full or half), standardized weaning weights of non-fostered siblings were not significantly asociated with fostered sibling weights (Fig. 3; R 2 = 0.03, d.f. = 1,44, P = 0.22). The slope of the estimated regresion line was b = -0.23 ? SE 0.18, which represents an estimate of heritability if pups were related by 100%. Growth rate of non-fostered siblings also did not explain a significant amount of the variation in fostered sibling growth rate (Fig. 4; R 2 = 0.10, d.f. = 1,33, P = 0.07; means ? 1 SE: non-fostered siblings: 3.60 g/day ? 0.11; fostered siblings: 3.54 g/day ? 0.14). The estimated slope of the line was b = -0.40 ? SE 0.21. Since both slopes are negative, resulting in a heritability estimate of 0 for both weaning weight and growth rate, it was not necesary to correct for the actual relatednes of S. columbianus siblings. DISCUSION Cross-fostering is a technique used to separate maternal and genetic contributions to phenotypic variation (Roff 1998). This study employed the cross-fostering method to determine if maternal efects are responsible for variation in offspring weaning weights and growth rates. The regresion of weaning weights of unrelated individuals (Fig. 1) raised in the same liter provides an estimate of the amount of variation in weight that is due to a common environment provided by the mother to the exclusion of direct genetic 15 variance. We found a highly significant asociation betwen foster and non-fostered weaning weights, suggesting that maternal efects contribute to ofspring weight. Murie et al. (1998) also found that weaning weight did not difer betwen unrelated litermates. The relationship betwen weaning weights of foster and non-fostered pups, however, could reflect an initial relationship betwen foster and non-fostered pups in weights at birth that persisted through the nursing period rather than maternal efects. However, no asociation was found betwen birth weights of foster pups and non- fostered pups. Therefore, during the period of maternal investment (e.g. lactation), foster pups and non-fostered pup weights became asociated, suggesting that maternal efects have a strong contribution to offspring weaning weights. Non-fostered pup growth rate also explained a significant amount of the variation in foster pup growth rate (Fig. 2), which suggests that the similarity of weights at weaning may be mediated through maternal contribution in the form of nutrient provisioning during lactation. Our results corroborate other studies that have shown significant maternal efects on the growth rate of body mas in rodents (Rutledge et al. 1972, Riska et al. 1984, McAdam et al. 2002). Growth rates and weaning weights were not identical betwen unrelated litermates as indicated by the significant diference betwen the estimated slopes of the regresion lines and the predicted slope of 1.0. This could be caused by initial diferences in non-foster and foster pup birth weights that persisted to weaning or by the ability of mothers to discriminate betwen their own and other young. However, foster and non-foster pups were similar in birth weights and birth weight was not corelated 16 with weaning weight for both non-fostered and foster pups. This suggests that the diferentiation betwen non-fostered pup and foster pup weaning weight and growth rate was not due to diferences in birth weight that persisted to weaning. Also, the deviations from the lines of slope 1.0 for both weaning weight and growth rate were not significantly diferent from zero, which indicates that mothers did not give preferential treatment to their own young. The lack of a relationship betwen whether a pup was fostered or not and survival to yearling age also suggests that mothers do not discriminate their own young from fostered young. This result is consistent with a two-year study by Murie et al. (1998), that found no diference in survival of juveniles that were fostered and juveniles that remained with the natal mother through the nursing period. The significant diference betwen the estimated regresion lines and a line of slope equal to 1.0 could be due to the use of ordinary least squares (OLS) regresion. When X and Y values are not independent the slope of the regresion line estimated by the OLS method tends to be underestimated (Gren 2001). Estimates of the amount of variation in ofspring phenotype due to maternal efects could potentialy be confounded by non-maternal environmental efects, such as diferences in temperature or climate. The time period in which maternal efects were examined in this study occurred during lactation, when pups remain underground in natal burrows. Liters could experience diferences in temperature due to diferences in the location and depth of natal burows. However, locations of natal burows are selected and the burrows created by the mother, and thus, could be considered a maternal efect. 17 The regresions of weaning weights and growth rates of unrelated litermates are based on the underlying asumption that prenatal maternal efects do not contribute to pup weight. A study by Moore et al. (1970) on mice showed that prenatal efects on pup weight were smal and did not persist past one wek following birth. However, Rhees et al. (1999) found that uterine maternal efects had significant and persistent contributions to offspring body weight and growth rate. In Richardson's ground squirels (Spermophilus richardsonii) mothers invest much more energy in pups during lactation than to developing embryos during gestation (Michener 1989). Asuming this also occurs in Columbian ground squirels, as it does in other rodent species (McClure 1987), the propensity for prenatal maternal efects on weaning body mas and growth rate are likely smal relative to postnatal maternal efects. Since pups were not always cross-fostered on the parturition date, some pups spent one to two days with the birth mother before cross-fostering occurred. Thus, relationships betwen unrelated litermates may be influenced by maternal investment by the natal mother prior to cross-fostering. However, if this were the case the result would be either a lack of a significant asociation betwen unrelated litermates or a weaker asociation than was actualy found. Thus, the actual relationship betwen fostered pups and the mother's own pups within a liter may be slightly stronger than was found in this study. Heritability of a character is estimated by comparing values of the trait betwen relatives. But, if relatives are raised in the same environment, the heritability estimate may be biased because it includes both environmental and genetic efects on the trait in 18 question (Smith and Wetermark 1995) . In our study, cross-fostering eliminates the shared environment betwen siblings during the energeticaly costly lactation period, and thus, alows for examination of genetic relatednes at the exclusion of potentialy confounding maternal efects. Of course, maternal efects during gestation are possible, so this estimate of heritability may stil include some environmental variation. Heritability estimates can also be biased by dominance when comparing traits of full- sibs. However, this would result in inflation of the heritability estimate, therefore estimates derived from regresions involving full-sibs are a maximum estimate (Falconer and Mackay 1996). In our study, heritability of juvenile weaning weight and growth rates was estimated as zero. Low heritabilities are often calculated from traits that have higher fitnes consequences (Falconer and Mackay 1996, Merila and Sheldon 2000). Although the relationships betwen weaning weights or growth rates and reproductive succes have not been examined in S. columbianus, the impact of weight on survival, a correlate of fitnes, has been explored. Weight atained by juveniles prior to hibernation has a significant impact on over-winter survival (Murie and Boag 1984) and over-winter survival of juveniles acounts for the majority of variation in a female's lifetime reproductive succes (King et al. 1991). Future research involving maternal efects on ofspring mas and growth rate should atempt to discern betwen environmental and genetic maternal efects. Since maternal efects are phenotypic contributions to offspring, they themselves consist of both environmental and genetic components (Krist 2004). An evolutionary response to 19 selection occurs only when the trait consists of heritable variation (Stearns 1992); thus, genetic maternal efects can have important evolutionary consequences. For example, selection can act on heritable maternal efects, which can alter the relationship betwen the additive genetic variation and phenotypic variation of a trait and result in changes in evolutionary rates (Wolf et al. 1998). 20 40 60 80 100 120 140 160 40 60 80 100 120 140 160 Non-fostered pup weaning weight (g) F o s t e r p u p w e a n i n g w e i g h t ( g ) Fig. 1 Relationship betwen weights of unrelated litermates at weaning. Liters are comprised of both foster pups and non-fostered pups. Foster pups are pups that were transfered from another liter and non-fostered pups refer to pups that remained with the birth mother. Weights were averaged within liters if more than one foster or non- fostered pup survived to weaning. Weight is measured in grams. A mother's own pup weight explained 64% of the variation in foster pup weight (P<0.0001). 21 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50 Non-fostered pup growth rate (g/day) F o s t e r p u p g r o w t h r a t e ( g / d a y ) Fig. 2 Plot of asociation betwen foster and non-fostered pup growth rates. Growth rates were calculated by subtracting weight at birth from weight at weaning and dividing by duration of the lactation period. Lactation duration is the number of days elapsed from birth to liter emergence. Foster pups are pups that were fostered into that liter. Non- fostered pups are pups that were not cross-fostered. Growth rates were averaged when there were multiple foster or non-fostered pups in a liter at weaning. The relationship betwen foster and non-fostered growth rates was positive and significant (R 2 = 0.58, P<0.0001). 22 -50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50 Non-fostered sibling weaning weight (g) F o s t e r e d s i b l i n g w e a n i n g w e i g h t ( g ) Fig. 3 Regresion of fostered sibling weaning weight on non-fostered sibling weaning weight. Non-fostered siblings are pups that remained with the natal mother. Fostered siblings are the siblings of the non-fostered pups that were fostered into another liter after birth. Weights were averaged if more than one fostered sibling or own sibling survived to weaning. Weaning weights standardized for liter size at weaning were used in the regresion, which were residuals obtained from ANOVA with liter size at weaning as the clas variable. Non-fostered sibling weaning weight only explained 3% of the variation in fostered sibling weaning weight (df = 1,44, P = 0.22). 23 -1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5 Non-fostered sibling growth rate (g/day) F o s t e r e d s i b l i n g g r o w t h r a t e ( g / d a y ) Fig. 4 Relationship betwen non-fostered sibling growth rate and fostered sibling growth rate. Non-fostered siblings are pups that remained with the natal mother. Fostered siblings are the siblings of non-fostered pups and were transfered into another liter after birth. Growth rates were calculated as the diference betwen the weaning weight and birth weight and dividing by length of the lactation period. Duration of lactation was estimated as the liter emergence date minus the liter birth date. Data standardized for liter size was used in the regresion and was obtained by extracting the residuals from an ANOVA using respective liter size at weaning as the grouping variable. Non-fostered sibling growth rate did not explain a significant amount of variation in fostered sibling growth rate (R 2 = 0.10, df = 1,33, P = 0.07). 24 CHAPTER TWO EFECTS OF MATERNAL CHARACTERISTICS ON OFSPRING WEIGHT, GROWTH RATE, AND SURVIVAL IN COLUMBIAN GROUND SQUIRELS Abstract. Ofspring phenotype is influenced by an individual?s own genetic composition, its environment and the phenotype of its mother. Therefore, variation in maternal characteristics may substantialy alter phenotypic expresion of her ofspring. In this study we tested possible relationships betwen maternal body condition, structural size, change in mas during reproduction, timing of reproduction, and liter size on offspring traits. Ofspring traits were weaning weight and rate of weight gain during the nursing period. To ases if maternal investment in ofspring traits extended beyond the period of maternal care, we looked for asociations betwen offspring and maternal traits and over- winter survival of pups and pup mas at spring emergence from hibernation the following year. Maternal variables explained a significant amount of variation in both average pup weaning weight (R 2 = 0.69) and growth rate (R 2 = 0.72) but not pup spring mas (R 2 = 0.34). Paterns of investment on growth rate and weaning weight were similar. Maternal body condition and liter size explained the most variation in pup growth rate (condition: p = 0.28; liter size: p = -0.88) and weaning weight (condition: p = 0.34; liter size: p = - 0.85). Pup growth rate had a positive efect on survival to yearling age (F [1, 11] = 4.57, P 25 = 0.03), whereas pup weaning weight and maternal characteristics had no efect on survival. 26 INTRODUCTION Maternal efects occur when a mother?s phenotype impacts the phenotype of her offspring (Bernardo 1996a). For example, a mother?s aces to high quality food resources may afect her condition, which can influence body mas or growth rate of her offspring. Growth rate and weight during early postnatal development are important because they are often related to survival, especialy through the first winter (Clutton- Brock et al. 1982, Murie and Boag 1984, C?t? and Festa-Bianchet 2001). Growth rate and weaning weight can also have persistent efects, as indicated by their correlation to adult body size (Clutton-Brock et al. 1982, Myers and Master 1983). Larger adult size can result in greater reproductive succes, possibly through stronger competitive abilities and diferential procurement of mates (Clutton-Brock et al. 1982). Variation in offspring phenotypes due to maternal efects can be caused by variation in maternal behaviors, such as parental care or timing of reproduction or through specific maternal traits that impact ofspring traits. Several factors, such as maternal size, parity, age, and timing of the reproductive season are known to influence offspring weight and growth rate in mamals. In many mamalian species maternal body mas is positively correlated with offspring mas (Skogland 1984, Hoogland 1995, C?t? and Festa-Bianchet 2001) and growth rate (Matingly and McClure 1982, Myers and Master 1983, Bowen et al. 2001), which could be atributed to the enhanced ability of larger females to acquire resources and provide food for their young (Matingly and McClure 1982). 27 In some mamalian species, offspring born earlier in the reproductive season have lower mas gain during lactation (Zwank and Zeno 1986) and timing of reproduction is negatively correlated with offspring mas (Fairbanks 1993, C?t? and Festa-Bianchet 2001). In ground-dweling sciurids, however, females that reproduce earlier often give birth to offspring that are lighter than later born individuals (Dobson and Michener 1995, Rieger 1996). Mating earlier may be advantageous because it alows females and juveniles more time to grow and acquire fat reserves before the onset of the winter months (Murie and Boag 1984, Milesi et al. 1999, C?t? and Festa-Bianchet 2001). Dobson et al. (1999) examined the influences of maternal body size and timing of reproduction on offspring mas, liter size, and liter mas in Columbian ground squirels, but found that maternal characteristics did not explain significant amounts of variation in average offspring mas. The lack of significance is surprising considering maternal efects constitute a large proportion of the variation in both offspring weaning mas and growth rate during lactation (Skibiel 2007). In addition, Dobson et al. (1999) did not find an asociation betwen offspring survival and pup weaning mas. The lack of significant paterns betwen maternal traits and offspring weight and betwen weight and survival may be atributed to the location of the population. The population studied occurs at a much lower elevation than the population in this study (695 m versus 1550 m in the current study) and diferences in life history traits among ground squirel populations at diferent elevations have been documented (Murie and Haris 1982, Dobson and Murie 1987, Dobson et al. 1992). In addition, Dobson et al. (1999) did not take number of 28 offspring into acount in the analyses, which could have influenced the results. Liter size has significant influences on offspring size (se results) and therefore should be controlled for statisticaly. Thus, further examination of the efects of maternal traits on offspring traits and survival in Columbian ground squirels is waranted. The first objective of this study was to test causal hypotheses of relationships betwen maternal size, change in mas during the reproductive season, and offspring weaning weight and growth rate, which was conducted through path analysis. Since mas can reflect both physiological condition and/or structural size (Dobson 1992, Dobson and Michener 1995, Dobson et al. 1999), body mas was partitioned into these two constituents. Body condition is a measure of the energetic composition of an organism in the form of muscle mas (Perins and McClery 2001), fat reserves, protein, and water (Dobson 1992, Schulte-Hostedde et al. 2005), whereas structural size represents appendage length or skeletal size (Dobson 1992). Since females of the same structural size can vary greatly in the amount of energy stores in the body, analyzing the influence of both maternal structural size and body condition may shed light on paterns of variation in reproductive output (Dobson and Michener 1995). Since mothers in beter condition, mothers with higher weight gain during reproduction, or larger mothers likely have more or beter quality resources available to them, they should invest more of the available energy into reproduction (Hirschfield and Tinkle 1975, Price 1998). Thus, we predicted that females in beter condition, larger females, and females with higher gain in mas during reproduction would raise ofspring with greater weaning weights and higher growth rates during lactation. We also expected 29 a positive influence of reproductive timing on ofspring weaning weight and growth rate. The second objective was to test the hypothesis that maternal influences on offspring traits persist to yearling age. This was acomplished by examining influences of maternal and offspring traits on survival to yearling age and on spring mas of pups at emergence from their first hibernation. We predicted that survival and spring emergence mas of pups would be positively asociated with offspring weaning weight, growth rate, maternal condition, mas gain, and structural size. We predicted a negative asociation betwen mating date and survival of ofspring and betwen mating date and spring emergence mas of pups the following year. MATERIALS AND METHODS Study location Field studies of Columbian ground squirels were conducted during the years 1992-2006. The population inhabited a meadow (110?W 50?N; elevation, 1550 m ) adjacent to Gorge Crek in the Sheep River Provincial Park, Alberta, Canada. Field methods Ground squirels were captured at spring emergence (April-May) by placing live- traps (Tomahawk #201 collapsible chipmunk trap, 16 x 5 x 5 cm 3 ) baited with peanut butter near burow openings. Al individuals were marked with numbered fingerling ear tags, weighed using a Pesola scale, and given unique body markings using hair dye (Lady Clairol Hydrience #52 black pearl). Zygomatic arch breadth was measured in 1993, 30 1994, and 2003-2006. In June and early July, juveniles were captured at emergence from natal burrows using the same techniques described above and also were marked with unique numbered ear tags. Time of liter emergence was estimated by adding 51 days to the mating date: 24 days from ating to parturition (Murie and Haris 1982) and 27 days from parturition to liter emergence (Murie 1992). Mating date was determined by examination of external reproductive morphology (Murie and Haris 1982) or by observations of pre-copulatory behavior, such as male interest in the female prior to but not following copulation, males entering burows with estrous females, or by observation of a copulatory plug. If a juvenile was captured the following spring after emergence from a hibernaculum it was recorded as surviving its first hibernation. In 1993, 1994, 1996, 1999, 2000, and 2003-2006, females that copulated were captured prior to parturition (May-June), weighed, and housed in the laboratory until birth (1-9 days). Females were kept in polycarbonate microvent rat cages (Alentown Caging Equipment Company; 267 x 483 x 203 m 3 ) on pine chip bedding and were supplemented with newspaper nesting material. To obscure vision from neighboring females and to simulate the burrow environment, cages were covered in vented black plastic bags and stored in a temperature controlled room. Squirels were fed a high protein horse fed (oats, barley, wheat, and compresed vegetable material in a molases mix; 13% crude protein) ad libitum and letuce and apple twice daily. Cages were checked for pups 3-4 times per day from 0530-2230 hrs. Females and neonates were weighed at least 4 hours after birth (to ensure that parturition was complete). Neonates were marked by removing a smal amount of tisue from the outer right or left toe bud on 31 a hind limb or from the tail. The nail does not grow on the toe where tisue was removed and a knot appears at the end of a tail that had tisue removed, alowing for identification of pups at emergence from the natal burows. Mothers were released on the meadow by opening the cage containing mother and pups and in the event that the mother did not retrieve the pups from the cage, pups were placed in the natal burow following imergence of the mother into that burrow. Variables Maternal characteristics examined included body condition, structural size, mas change during reproduction and timing of reproduction. Both body mas and zygomatic arch breadth were measured at first capture after spring emergence from hibernation. Since zygomatic arch breadth is asociated with other skeletal measurements (Dobson et al. 1999), zygomatic arch breadth was used as an estimate of structural size. In this study females mated, on average, 3.66 days following first capture after spring emergence; therefore measures of mas and zygomatic arch breadth approximate size at the beginning of the gestation period. Body condition was estimated from the residuals of the regresion of spring emergence mas on zygomatic arch breadth. Change in mas over the reproduction period is the diference betwen female's body mas at spring emergence and her mas at the time of pup emergence from the nest burrow. Emergence of pups from nest burows is an indication of liter weaning and cesation of lactation (Anderson et al. 1976, Murie and Dobson 1987, Michener 1989, Rieger 1996). Timing of reproduction is the julian date that mating occurred. Ofspring traits included weaning 32 weight and growth rate during the nursing period, and pup spring emergence mas the following year. Growth rate was calculated by subtracting weight at birth from weight at weaning and dividing by the length of the lactation period. When more than one pup within a liter survived to weaning, weaning weight and growth rate where averaged within liters. Spring emergence mas of pups is the weight at first capture of ofspring following their first hibernation. For analysis of the influence of maternal variables on pup spring emergence mas, mas of pups was averaged within liters when more than one pup within a liter survived to yearling age. Multiple records for some females were considered independent because 1) other studies show that reproduction in one year does not influence reproduction in the following year (Risch et al. 1995, Dobson et al. 1999, Broussard et al. 2005b) and 2) weaning weight in one year was not asociated with weaning weight the following year, liter size in one year was not correlated to liter size the next year, and growth rate in one year was not asociated with growth rate the following year (Spearman rank correlation; respectively, r = 0.17, n = 92, P = 0.10; r = 0.06, n = 92, P = 0.59; r = -0.07, n = 22, P = 0.77). Statistical analysis Al analyses were conducted using SAS statistical software for Windows (SAS Institute 2002). Tests employed included general linear models (PROC GLM, used for analysis of variance [ANOVA]), correlations (PROC OR), mixed models (PROC MIXED used for logistic regresion) and path analysis (STB option of PROC REG). VIF 33 and COLIN options of PROC REG were utilized to detect biases due to co-linearity of independent variables. Significance level for al tests was ! = 0.05. Data were asesed for normality graphicaly or by performing Shapiro-Wilks tests. General linear models and Tukey tests for multiple comparisons of means were used to explore diferences in maternal characteristics and offspring traits among females of diferent ages. Yearlings were not included in analyses because only 9 yearlings during the 15 year study raised a liter to weaning and only 2 of those yearlings exhibited characteristics necesary to conduct the path analyses. Yearlings were not pooled with other age groups because studies have found diferences in life history traits betwen yearlings and older females in Columbian ground squirels (Dobson et al. 1999, Broussard et al. 2005a). Females aged 7-10 were pooled due to smal sample sizes. Path analysis was used to determine the interelationships among maternal and offspring traits (Fig. 1). Since tradeoffs betwen liter size and offspring traits were evident (Table 2), liter size was controlled for statisticaly by including it as an independent variable in the path analyses. Path analysis (Li 1981) involves first diagraming a hypothetical model, a priori, that incorporates causal relationships among the variables. The null hypothesis of no causal relationship is tested by the significance of path coeficients. Path coeficients are standardized partial regresion coeficients which indicate the magnitude of the efect of the independent variable on the dependent variable, while al other variables are held statisticaly invariant (Li 1981, Pedhazur 1982, Sokal and Rohlf 1995). Standardized path coeficients alow for the inclusion of 34 variables measured on diferent scales, such as mas (in grams) and timing of reproduction (in days). As in al multiple regresion techniques, co-linearity of independent variables in path analysis can result in eroneous conclusions. Co-linearity of independent variables can inflate standard erors of path coeficients, increasing the probability of a type I eror, and can also inflate values of the path coeficients. Inflation of standard erors due to co-linearity can be determined by examining variance inflation factors (VIFs) (Petraitis et al. 1996). VIFs greater than 10 indicate biases of standard erors due to collinearity of independent variables (Myers 1990). Biases in values of path coeficients can be detected through condition indices. Condition indices are derived from the eigenvalues of the correlation matrices of independent variables (Petraitis et al. 1996). Condition indexes betwen 5 and 10 suggest weak biases due to co-linearity while condition indexes betwen 30 and 100 indicate strong biases (Belsley et al. 1980). As in Dobson et al. (1999), years were subdivided acording to timing of the reproductive season and path analyses were conducted separately for years of early and years of late reproduction, adjusting variables for efects of years among early years and among late years. Years were categorized as early (years 1992, 1994-2000, 2005) if the mean date of mating for that year was before the mean date of mating for al years in the study. Late years (years 1993, 2001-2003, 2005, 2006) were years in which mean mating date was after the mean mating date of the entire study. Mean Julian date for early years was 117 and mean date for late years was 122. Since paterns of maternal investment did not difer betwen early and late years, al years were combined to perform path analyses. 35 Diferences in maternal and offspring characteristics were first examined for variation among years in the combined data set using general linear models. Possible relationships betwen offspring growth rate and weaning weight and spring emergence mas of pups were examined by conducting Spearman rank correlations. When more than one pup in a liter survived to yearling age (n = 27 liters out of 116 liters with surviving pups), one pup was randomly selected for inclusion in the analysis. This resulted in removal of 31 pups out of 109 pups with data on spring emergence mas. Mixed model logistic regresion was used to ases influences of maternal traits and offspring growth rate and weaning weight on survival to yearling age. Year and liter identification number (liter ID) were included in the model as random variables. This statisticaly removed the efect of the year on offspring survival and controlled for non- independent data, due to use of multiple pups within liters. Liter ID was a number arbitrarily asigned to each liter. RESULTS Spring mas, date of mating, and zygomatic arch breadth difered significantly among females of diferent ages, while mas change during reproduction, body condition, liter size at weaning, average weaning weight, average growth rate of offspring, and average pup spring emergence mas did not. On average, two year olds were lighter at spring emergence, had later mating dates, and had narower zygomatic arch breadths than females in other age clases (Table 1). Tukey?s test also revealed a significant diference 36 in zygomatic arch breadth betwen the 3-year old and 6-year old age clases. Because females older than 2-years did not difer in any maternal or offspring traits (except zygomatic arch breadth, above) and because 2-year olds did not difer from older females in any of the dependent variables (viz. offspring traits) used in the path analyses, females of al age clases were pooled for further analyses. Significant variation among years was detected for mating date (F [13, 146] = 5.87, P < 0.0001), mother?s mas near the time of pup weaning (F [1, 9] = 3.88, P < 0.0001), zygomatic arch breadth (F [7, 10] = 3.72, P = 0.001), and maternal spring mas (F [13, 160] = 1.89, P = 0.03), but not for mother?s body condition at spring emergence (F [7, 109] = 1.04, P = 0.41), liter size at weaning (F [12, 92] = 0.76, P = 0.69), pup weaning weight (F [12, 92] = 1.09, P = 0.38), offspring growth rate (F [5, 52] = 0.43, P = 0.82), or spring emergence mas of pups the following year (F [8, 35] = 1.67, P = 0.14). Therefore, spring mas, mating date, mother?s mas at pup weaning, and zygomatic arch breadth were standardized for years by using residuals extracted from the ANOVA, where year was the grouping factor. Mas change during reproduction was calculated by subtracting mother?s spring mas from mother?s mas at pup weaning, both of which were adjusted for year efects. Of the females that raised a liter to weaning, 99% gained mas from spring emergence to the time of pup weaning (n = 248). Al variables were normaly distributed, except average pup growth rate, which was log transformed for subsequent analyses. Physiological condition of females was estimated by extracting residuals from the regresion of spring mas on zygomatic arch breadth. Spring mas increased with 37 zygomatic arch breadth (R 2 = 0.41, n = 118, P < 0.0001). Residuals were normaly distributed (W = 0.99, n = 118, P = 0.30). Therefore, residuals were used as an index of body condition in analyses. Maternal traits explained 69% of the variation in ofspring weaning weight, which was significant (n = 66, P <0.0001). A female?s body condition and liter size had significant efects on ofspring weaning weight (condition: p = 0.34, P = 0.0007; liter size: p = -0.85, P <0.0001; Fig. 2a). Efects of mating date and structural size on weaning weight approached significance (mating date: p = 0.14, P = 0.07; structural size: p = 0.14, P = 0.07). Similar paterns were evident for maternal efects on pup growth rate. Maternal traits explained a high and significant amount of variation in average pup growth rate (R 2 = 0.72, n = 55, P < 0.0001). Maternal body condition at spring emergence and liter size contributed significantly to variation in offspring growth rate (respectively; p = 0.28, P = 0.009; p = -0.88, P < 0.0001; Fig. 2b). There was a trend for a positive efect of mating date and structural size on growth rate, but these paths were not significant (mating date: p = 0.14, P = 0.07; structural size: p = 0.15, P = 0.07). Maternal traits did not explain a significant amount of variation in average pup spring mas (R 2 = 0.34, n = 28, P = 0.08), although sample size was smal. Due to lack of significance, results of the path analysis are not presented. In addition, there was a lack of an asociation betwen a pup?s weaning weight and its spring emergence mas (r = 0.16, n = 87, P = 0.14) and betwen a pup?s growth rate and its mas at spring emergence (r = 0.33, n = 32, P = 0.07). 38 Of the independent variables included in the path analysis, body condition and change in mas were highly negatively corelated (Table 2, Fig. 2). Females in beter condition at spring emergence had lower mas gain during reproduction. Growth rate and weaning weight were also highly negatively correlated with liter size, growth rate was highly positively correlated with weaning weight, and spring emergence of pups from hibernation was positively asociated with growth rate (Table 2). Despite inter-correlations among independent variables, co-linearity was not likely a cause of strong bias in estimation of path coeficients or their standard erors. The maximum VIF for the path analysis of weaning weight was 1.84, for growth rate was 1.95, and for pup spring mas was 1.49, which are wel under the value of 10 recommended by Petraitis et al. (1996). In addition, the highest condition indices for path models were 6.47, 6.43, and 6.94 (weaning weight, growth rate, and pup spring mas, respectively). Condition indices ranging from 5 to 10 indicate slight inflation of path coeficients while condition indices greater than 30 suggest high inflation of path coeficients (Belsley et al. 1980). Pup growth rate during the nursing period had a positive efect on survival (F [1, 11] = 4.57, P = 0.03). However, pup weaning weight did not afect survival to yearling age (F [1, 247] = 3.54, P = 0.06). Neither liter size at weaning nor any of the maternal traits had an efect on pup survival (liter size: F [1, 136] = 1.22, P = 0.27; mating date: F [1, 136] = 2.93, P = 0.09; structural size: F [1, 136] = 1.41, P = 0.24; mas change during reproduction: F [1, 136] = 1.18, P = 0.28; condition: F [1, 136] = 0.07, P = 0.79). 39 DISCUSION Before testing causal relationships of maternal traits on offspring traits, age- specific diferences in maternal variables were evaluated. Unlike other studies of Columbian ground squirels that only found diferences in maternal traits betwen yearlings and older females (Dobson et al. 1999, Broussard et al. 2003), we found diferences betwen 2-year olds and older females in structural size, timing of reproduction, and spring emergence mas. Populations of Columbian ground squirels inhabiting higher latitudes and higher elevations consist of few breding yearlings (Dobson and Murie 1987). Thus, 2 year olds are likely primiparous females, which tend to be lighter at spring emergence and emerge later from hibernation (Table 1). Path analysis resulted in diferent conclusions regarding influences of maternal traits on offspring mas and growth rate than did simple correlations. In correlation analysis maternal body condition at spring emergence was not asociated with offspring weaning weight or growth rate but was significant in path analyses. Significant efects appearing in path analyses but not correlation is likely due to the path analysis testing individual pathways by holding other paths statisticaly invariant. This controls for non- significant correlations among independent variables, which otherwise might mask significant paths (Dobson et al. 1999). Maternal characteristics explained a significant amount of variation in pup weaning weight and growth rate. Path analysis revealed a significant positive efect of a mother?s body condition on both weaning weight and growth rate (Fig. 2). This corroborates findings in a study of Richardson's ground squirels where a positive 40 correlation was found betwen average offspring mas and maternal body condition (Dobson and Michener 1995). As expected in the path analyses, liter size was significantly negatively correlated with both weaning weight and growth rate. In many mamalian species tradeoffs exist betwen offspring size or growth and number, such that liter size increases at the expense of individual size or growth (Cameron 1973, Michener 1989, Rieger 1996). The lack of significant paths from structural size and change in mas to weaning weight and growth rate was somewhat surprising. In Richardson?s ground squirels, maternal mas gain from estrus to birth had positive efects on neonate mas (Dobson and Michener 1995) and in Columbian ground squirels both yearling and older mothers that gained more weight during reproduction produced larger liters (Dobson et al. 1999). Females that gained more weight during reproduction were expected to raise larger individual pups with higher growth rates because the additional energy acquired could be invested in the liter. Larger females were expected to produce larger pups with higher mas gain, possibly reflecting beter competitive ability to obtain higher quality teritories (Murie and Haris 1988, Boag and Wigget 1994). It appears, however, that body condition is more important in determining pup weaning weight and growth rate than is gain in mas or structural size. Timing of reproduction has been shown to influence offspring mas in several species of ground dweling sciurids. In both Richardson's ground squirels (Dobson and Michener 1995) and Uinta ground squirels (Rieger 1996), earlier reproduction resulted in lighter offspring but larger liters. A seasonal decline in liter size, such that as the 41 active season progreses liter size becomes smaler, with a concomitant increase in individual offspring size as the season progreses, was expected in this population of Columbian ground squirels. However, we found a lack of a significant relationship betwen mating date and offspring mas and growth rate. In another study of Columbian ground squirels, reproductive timing was not found to significantly influence liter size when data from al years of the study were combined (Dobson et al. 1999). However, when data were sorted by timing of the reproductive season, a significant relationship betwen mating date and liter size in years of early breding was revealed. In this population of Columbian ground squirels, there was no diference in paterns of investment when years were categorized by the timing of the breding season. Thus, there appears to be no efect of reproductive seasonality, either timing of breding within a season or timing of the breding season among years, on offspring weaning weight or growth rate. While standard erors of path coeficients were not likely biased in the path analyses [variance inflation factors were within the range suggested by Petraitis et al. (1996)], the highest condition indices for both path analyses were around 6, suggesting slight inflation of path coeficients. High condition indices were for paths from maternal change in mas during the reproductive season to ofspring weaning weight and growth rate. Biases due to co-linearity should have elevated path coeficients for these paths, which was opposite the evident patern. Since path coeficients were stil not significant, despite possible inflation, conclusions regarding the efect of maternal mas change on offspring traits were not influenced by this bias. 42 Maternal traits did not explain a significant amount of the variation in mas of pups at spring emergence from hibernation. However, this could be atributed to the smal sample size. Ofspring growth rate and weaning weight were also not asociated with pup spring mas. These results suggest that maternal influences on growth rates and weaning weights of their young disipate over time. This opposes the findings for red deer and deer mice that juvenile body size or growth is correlated to adult body size (Clutton-Brock et al. 1982, Myers and Master 1983). Although offspring traits were not asociated with mas of the pup at spring emergence, they may be indicators of survival. Pups that grew faster during the nursing period survived more than slower growing pups. Pup weaning weight also showed a positive trend, whereby heavier pups had a beter chance of survival to yearling age, although this asociation was not significant. Unlike the lower elevation population studied by Dobson et al. (1999), we found that liter size and maternal characteristics did not have an efect on the over-winter survival of pups. In this population of Columbian ground squirels, mothers invest in their offspring by altering growth rates and weaning weights of their young. Particularly, mothers in beter condition at spring emergence from hibernation produce pups that grow faster during the period of maternal care and weigh more at weaning than pups born from mothers in poor body condition. We also found that efects of maternal investment on offspring growth rate and weaning weight do not appear to persist to yearling age, in that they are not asociated with pup mas at emergence from hibernation. However, growth rate appears to be especialy important because of its influence on offspring survival. 43 Pups that grow faster during the nursing period survive more than pups that grow at a slower rate. These results are diferent from the maternal investment paterns described by Dobson et al. (1999). Dobson et al. (1999) found that both yearling and older females alter reproductive output by producing larger liters rather than producing heavier young. None of the maternal characteristics we examined were correlated with liter size at weaning, suggesting that mothers do not increase reproductive output by altering liter size. In addition, Dobson et al. (1999) suggested that liter size was the best predictor of survival to yearling age whereas our results suggest that offspring growth rate is the best predictor of survival. It is curently unclear if diferences in maternal investment paterns betwen the two populations are due to diferences in location of the populations or due to diferences in treatment of variables in statistical analyses. Further research is necesary to determine cause of diferences in maternal investment paterns among populations of Columbian ground squirels. 44 TABLE 1. Averages of maternal characteristics and offspring traits for each age clas ? 1 standard eror of the mean. Sample sizes are in parentheses. Females age 7-10 were pooled due to low sample sizes. Mas change is the diference betwen spring emergence mas and mas at liter emergence. Body condition was estimated from residuals of the regresion of mas at spring emergence on zygomatic arch breadth. Growth rate was calculated by subtracting birth weight from weaning weight and dividing by the duration of the lactation period. F-values, degres of fredom, and P-values correspond to results of ANOVAs testing diferences in variables among age clases. P-values were significant at ! = 0.05. Age clas (years) Variable 2 3 4 5 6 7-10 F d.f. P Spring mas* (g) 389 ? 4.7 (79) 425 ? 5.1 (56) 434 ? 6.4 (48) 442 ? 9.7 (48) 443 ? 9.3 (27) 445 ? 7.4 (39) 16.2 5, 291 <0.0001 Mating date* 122 ? 0.7 (67) 119 ? 0.6 (55) 119 ? 1.0 (48) 118 ? 0.8 (47) 117 ? 1.0 (28) 117 ? 1.1 (38) 4.9 5, 277 0.0003 Mas change 86 ? 7.5 (41) 81 ? 8.8 (39) 73 ? 8.5 (33) 110 ? 8.6 (40) 83 ? 22.7 (26) 97 ? 12.8 (30) 1.5 5, 203 0.19 45 Structural size* (m) 33.2 ? 0.16 (35) 33.9 ? 0.16 (19) 34.2 ? 0.11 (21) 34.4 ? 0.14 (21) 34.9 ? 0.19 (7) 34.5 ? 0.16 (15) 13.8 5, 112 <0.0001 Body condition -0.20 ? 6.5 (35) 6.33 ? 7.88 (19) 3.20 ? 6.91 (20) -0.32 ? 7.33 (21) -16.03 ? 15.92 (7) -8.11 ? 12.35 (15) 0.53 5, 111 0.75 Liter size 2.63 ? 0.11 (41) 2.83 ? 0.16 (35) 2.75 ? 0.16 (32) 2.86 ? 0.16 (37) 3.33 ? 0.25 (24) 3.08 ? 0.22 (25) 2.0 5, 188 0.08 Weaning weight (g) 109 ? 3.9 (41) 110 ? 3.4 (35) 114 ? 3.1 (32) 116 ? 3.4 (36) 109 ? 3.7 (24) 109 ? 5.4 (25) 0.7 5, 187 0.63 Growth rate (g/day) 3.81 ? 0.21 (18) 3.82 ? 0.33 (11) 4.34 ? 0.24 (9) 3.85 ? 0.20 (17) 3.56 ? 0.18 (9) 3.85 ? 0.31 (13) 0.7 5, 71 0.60 Pup spring mas (g) 245 ? 6.3 (13) 270 ? 7.1 (21) 255 ? 7.3 (16) 257 ? 8.9 (19) 255 ? 10.6 (10) 265 ? 13.2 (10) 1.03 5, 83 0.41 *Significant diferences betwen 2-year olds and al other age clases (Tukey multiple range tests). Tukey?s test also showed a significant diference betwen 3-year olds and 6-year olds in structural size. 46 TABLE 2. Correlations of variables used in path analyses. P-values are located under the correlation coeficients. Sample size is in parentheses. P-values les than 0.05 were considered significant. Mas change is the diference betwen spring emergence mas and mas at liter emergence. Body condition was estimated from residuals of the regresion of mas at spring emergence on zygomatic arch breadth. Growth rate is the gain in mas during lactation divided by lactation length. Structural size, mating date, and mas change are year standardized. Growth rate is log transformed. Mating date Mas change Zygomatic Breadth (m) Wean weight (g) Growth rate (g/day) Liter size Pup spring mas (g) Condition -0.02 0.81 (110) -0.62 <0.0001 (66) 0.01 0.88 (117) 0.1 0.35 (70) 0. 21 0.38 (55) 0.12 0.31 (70) -0.10 0.58 (30) Mating date - 0.14 0.17 (107) -0.16 0.09 (111) 0.01 0.90 (105) 0.04 0.74 (58) 0.001 0.99 (105) -0.21 0.17 (44) Mas change - 0.15 0.22 (66) -0.08 0.42 (101) -0.17 0.20 (58) 0.17 0.10 (101) -0.17 0.27 (42) Zygomatic breadth - 0.04 0.74 (70) 0.006 0.96 (55) 0.18 0.13 (70) 0.004 0.98 (30) Weaning weight - 0.96 <0.0001 (58) -0.65 <0.0001 (105) 0.25 0.10 (44) 47 Growth rate - -0.80 <0.0001 (58) 0.46 0.04 (21) Liter size - -0.24 0.12 (44) 48 FIG. 1. Path model of the efects of maternal characteristics and liter size on offspring traits. Single headed-arows represent paths from independent variables to the dependent variable. Double headed-arows represent correlations betwen independent variables. Mas change is the diference betwen spring emergence mas and mas at liter emergence. Body condition was estimated from residuals of the regresion of mas at spring emergence on zygomatic arch breadth. Ofspring traits are average pup weaning weight, average pup growth rate, and average pup mas at spring emergence from hibernation. U is the path coeficient for unexplained variation in offspring traits and was calculated by the equation 2 1!(Li 1981) Ofspring trait Structural size Mas change Body condition Mating date U Liter size 49 A. B. FIG. 2. Path models for the influence of maternal characteristics on (A) ofspring weaning weight (n = 66) (B) ofspring growth rate (n = 55). Mas change is the diference betwen spring emergence mas and mas at liter emergence, both adjusted for among year variation. Body condition was estimated from residuals of the regresion of mas at spring emergence on zygomatic arch breadth. Structural size and mating date are adjusted for year efects. Growth rate is log transformed. Path coeficients are given Weaning weight Structural size Mas change Body condition Mating date U Liter size -0.62 0.34 0.14 0.14 0.14 -0.85 0.56 Growth rate Structural size Mas change Body condition Mating date U Liter size -0.62 0.28 0.14 0.09 0.15 -0.88 0.53 50 to the right of maternal characteristics. Mas change and body condition are significantly correlated and the corelation coeficient is located to the left of maternal characteristics. U is the path coeficient for unexplained variation in offspring traits and was calculated by the equation 2 1!(Li 1981). Solid lines are positive paths and dashed lines are negative paths. Bold arows indicate significant path coeficients. 51 CONCLUSIONS Maternal efects are ubiquitous in nature, occurring in many diferent species of both plants and animals. Maternal efects are a type of environmental influence that can modify offspring phenotypes and might occur through a variety of mechanisms, including directly through maternal traits, through transmision of antibodies or hormones, or through maternal behaviors, such as maternal care (Mather and Jinks 1971, Mousseau and Dingle 1991, Clark and Galef 1995). The current study addresed four questions regarding maternal efects in Columbian ground squirels: 1) Are maternal efects responsible for variation in ofspring phenotypes? 2) How do paterns of maternal investment influence offspring phenotypes? 3) Do influences of maternal characteristics on offspring phenotype disipate over time? 4) Do maternal and offspring traits afect offspring survival to yearling age? The first chapter atempted to determine if maternal efects contributed to variation in ofspring growth rate and weaning weight. For approximately one month following birth, Columbian ground squirel pups remain underground in natal burows, where their only source of nutrition is through the mother?s milk. This provides ample opportunities for maternal efects to occur. Therefore, we expected development of pups during the nursing period to be strongly influenced by maternal efects. A cross-fostering design was used to disentangle influences of direct genetic efects (i.e. heritability) from maternal efects on ofspring development. In a cross-fostering experiment, liters are 52 paired at birth and pups are swapped betwen the liters. Thus, each liter consists of the mother?s own pups and unrelated pups that were fostered into the liter. Non-fostered pups explained 65% and 58% of the variation in weaning weight and growth rate of their unrelated litermates (e.g. fostered pups), respectively. However, some diferentiation in weaning weight betwen unrelated litermates occurred because the slope of the estimated regresion line was significantly diferent from 1.0. This suggested that there were either initial diferences in weight betwen unrelated litermates at birth, before cross-fostering, or that mothers were favoring their own young. However, further analysis revealed that unrelated litermates were not diferent in birth weight and that mothers did not invest more in their own birth pups. In addition, there was no asociation betwen survival to yearling age and type of pup (viz., foster or non-fostered). Thus, the diferentiation betwen weaning weights of unrelated litermates was likely an artifact of the type of regresion used (Gren 2001), rather than to diferential investments by mothers to their own pups. Genetic siblings that were raised in diferent liters were not similar in either growth rate or weaning weight, resulting in an estimate of heritability of 0. These results indicate that maternal efects were more important in contributing to variation in pup development than were heritable genetic efects. These results are consistent with a study of maternal efects on growth rate and weight in another sciurid species that found low heritability and high contribution of maternal efects (McAdam et al. 2002). Chapter two sought to examine influences of specific maternal phenotypes, or traits, on offspring weaning weight and growth rate and the persistence of these efects. 53 Maternal traits investigated included timing of reproduction, body condition, structural size, and change in mas during reproduction. These maternal traits have been shown to influence offspring growth rate and weaning weight in other mamalian species (Zwank and Zeno 1986, Fairbanks 1993, Rieger 1996, Bowen et al. 2001, C?t? and Festa- Bianchet 2001). In addition, empirical evidence suggests that both growth rate and weaning weight are important for survival, especialy to yearling or reproductive age (Clutton-Brock et al. 1982, Murie and Boag 1984, C?t? and Festa-Bianchet 2001). We found that the only maternal trait to influence growth rate and weaning weight was a female?s body condition at spring emergence from hibernation. This finding suggests that body condition is the only maternal trait important in the variation of pup growth rate and weaning weight. Liter size also had a significant influence on growth rate and weaning weight, which was not surprising given the abundance of species that exhibit tradeoffs betwen liter size and growth rate or weaning weight (Cameron 1973, Michener 1989, Rieger 1996). We also found that only growth rate was asociated with survival of pups to yearling age, although weaning weight exhibited a positive, non- significant asociation with survival. Liter size at weaning and none of the maternal traits investigated were correlated with pup survival. These results difer from findings of a similar analysis on Columbian ground squirels inhabiting an area of lower elevation (Dobson et al. 1999). In that study, the maternal traits examined (which were the same as in this study) did not have an influence on offspring weight. However, in older females al maternal traits, except for timing of reproduction, had a positive influence on both liter size and total liter mas at weaning. 54 Dobson et al. (1999) also found that liter size was the best predictor of the number of surviving pups. In the present study, we found that growth rate had a significant positive efect on ofspring survival. Diferences betwen the two studies on Columbian ground squirels suggest that paterns of maternal investment may not be general to the whole species, but rather vary across populations and perhaps even among years. In fact, diferences may be atributable to diferences in elevations among habitats for diferent populations. Studies on Columbian ground squirels indicate that diferences in life histories occur among diferent populations at diferent elevations. Specificaly populations occurring at higher elevations have later breding dates, lower body mas and poorer body condition, smaler liters, higher survival, and females mature at a later age (Murie and Haris 1982, Dobson and Murie 1987, Dobson 1992). More research is necesary to determine if diferences in investment paterns by mothers occur among the diferent elevations at which populations occur. 55 LITERATURE CITED Anderson, D. C., K. B. Armitage, and R. S. Hoffmann. 1976. Socioecology of marmots: female reproductive strategies. Ecology 57:552-560. Belsley, D. A., E. Kuh, and R. E. Welsch. 1980. Regresion diagnostics: identifying influential data and sources of collinearity. John Wiley & Sons, Inc., New York. Bernardo, J. 1996a. Maternal efects in animal ecology. American Zoologist 36:83-105. Bernardo, J. 1996b. The particular maternal efect of propagule size, especialy egg size: paterns, models, quality of evidence and interpretation. American Zoologist 36:216-236. Blums, P., R. G. Clark, and A. Mednis. 2002. Paterns of reproductive efort and succes in birds: path analyses of long-term data from European ducks. Journal of Animal Ecology 71:280-295. Boag, D. A., and D. R. Wigget. 1994. Food and space: resources defended by teritorial parous female Columbian ground squirels. Canadian Journal of Zoology 72:1908-1914. Bowen, W. D., S. L. Elis, S. J. Iverson, and D. J. Bones. 2001. Maternal efects on offspring growth rate and weaning mas in harbour seals. Canadian Journal of Zoology 79:1088-1101. 56 Broussard, D. R., F. S. Dobson, and J. O. Murie. 2005a. The efects of capital on an income breder: evidence from female Columbian ground squirels Canadian Journal of Zoology 83:546-552. Broussard, D. R., G. R. Michener, T. S. Risch, and F. S. Dobson. 2005b. Somatic senescence: evidence from female Richardson's ground squirels. Oikos 108:591- 601. Broussard, D. R., T. S. Risch, F. S. Dobson, J. O. Murie, and 2003. Senescence and age- related reproduction of female Columbian ground squirels. Journal of Animal Ecology 72:212-219. Bull, J. J. 1987. Evolution of phenotypic variance. Evolution 41:303-315. Cameron, G. N. 1973. Efect of liter size on postnatal growth and survival in the desert woodrat. Journal of Mamalogy 54:489-493. Campbel, M. T., and N. A. Slade. 1995. The efect of maternal mas on liter size and offspring survival in the hispid cotton rat (Sigmodon hispidus). Canadian Journal of Zoology 73:133-140. Clark, M. M., and B. G. J. Galef. 1995. Prenatal influences on reproductive life history strategies. Trends in Ecology and Evolution 10:151-153. Clutton-Brock, T. H. 1991. The evolution of parental care. Princeton University Pres, Princeton, New Jersey. Clutton-Brock, T. H., F. E. Guinnes, and S. D. Albon. 1982. Red deer; behavior and ecology of two sexes. University of Chicago Pres, Chicago. 57 C?t?, S. D., and M. Festa-Bianchet. 2001. Birthdate, mas and survival in mountain goat kids: efects of maternal characteristics on forage quality. Oecologia 127:230-238. Crespi, E. J., and H. Lesig. 2004. Mothers influence offspring body size through post- oviposition maternal efects in the redbacked salamander, Plethodon cinereus. Behavioural ecology 138:306-311. Dobson, F. S. 1992. Body mas, structural size, and life-history paterns of the Columbian ground squirel. American Naturalist 140:109-125. Dobson, F. S. 1995. Regulation of population size: Evidence from Columbian ground squirels. Oecologia 102:44-51. Dobson, F. S., M. J. Badry, and C. Geddes. 1992. Seasonal activity and body mas of columbian ground squirels. Canadian Journal of Zoology 70:1364-1368. Dobson, F. S., and G. R. Michener. 1995. Maternal traits and reproduction in Richardson's ground squirels. Ecology 76:851-862. Dobson, F. S., and J. O. Murie. 1987. Interpretation of intraspecific life history paterns evidence from Columbian ground squirels. American Naturalist 129:382-397. Dobson, F. S., and M. K. Oli. 2001. The demographic basis of population regulation in Columbian ground squirels. American Naturalist 158:236-247. Dobson, F. S., T. S. Risch, and J. O. Murie. 1999. Increasing returns in the life history of Columbian ground squirels. Journal of Animal Ecology 68:73-86. Elis, S. L., W. D. Bowen, and D. J. Bones. 2000. Maternal efects on offspring mas and stage of development at birth in the harbor seal, Phoca vitulina. Journal of Mamalogy 81:1143-1156. 58 Fairbanks, W. S. 1993. Birthdate, birthweight, and survival in pronghorn fawns. Journal of Mamalogy 74:129-135. Falconer, D. S., and T. F. C. Mackay. 1996. Introduction to Quantitative Genetics, 4th edition. Prentice Hal, England. Festa-Bianchet, M., and W. J. King. 1991. Efects of liter size and population dynamics on juvenile and maternal survival in Columbian ground squirels. Journal of Animal Ecology 60:1077-1090 Gren, A. J. 2001. Mas/length residuals: measures of body condition or generators of spurious results. Ecology 82:1473-1483. Hare, J. F., and J. O. Murie. 1992. Manipulation of liter size reveals no cost of reproduction in Columbian ground squirels. Journal of Mamalogy 73:449-454. Heath, D. D., and D. M. Blouw. 1998. Are maternal efects in fish adaptive or merely physiological side efects? Pages 178-201 in T. A. Mousseau and C. W. Fox, editors. Maternal efects as adaptations. Oxford University Pres, New York. Hirschfield, M. F., and D. W. Tinkle. 1975. Natural selection and evolution of reproductive efort. Procedings of the National Academy of Sciences 72:2227- 2231. Hoogland, J. L. 1995. The black-tailed prairie dog. University of Chicago Pres, Chicago. Kaplan, R. H. 1998. Maternal efects, developmental plasticity, and life history evolution: an amphibian model. Pages 244-260 in T. A. Mousseau and C. W. Fox, editors. Maternal efects as adaptations. Oxford University Pres, New York. 59 King, W. J., M. Festa-Bianchet, and S. E. Hatfield. 1991. Determinants of reproductive succes in female Columbian ground squirels. Oecologia 86:528-534. Kolbe, J. J., and F. J. Janzen. 2001. The influence of propagule size and maternal nest- site selection on survival and behaviour of neonate turtles. Functional Ecology 15:772-781. Krist, M. 2004. Maternal efects and ofspring performance: in search of the best method. Oikos 106:422-426. Kyneb, A., and S. Toft. 2006. Efects of maternal diet quality on offspring performance in the rove beetle Tachyporus hypnorum. Ecological Entomology 31:322-330. Lacey, E. P. 1998. What is an adaptive environmentaly induced parental efect? Pages 54-66 in T. A. Mousseau and C. W. Fox, editors. Maternal efects as adaptations. Oxford University Pres, New York. Li, C. C. 1981. Path analysis - a primer. Boxwood, Pacific Grove, California. Mather, K., and J. L. Jinks. 1971. Biometrical genetics. Chapman and Hal, New York. Matingly, D. K., and P. A. McClure. 1982. Energetics of reproduction in large-litered cotton rats (Sigmodon hispidus). Ecology 63:183-195. McAdam, A. G., S. Boutin, D. Reale, and D. Berteaux. 2002. Maternal efects and the potential for evolution in a natural population of animals. Evolution 56:846-851. McClure, P. A. 1987. The energetics of reproduction and life histories of cricetine rodents (Neotoma floridana and Sigmodon hispidus). Symposium of the Zoological Society of London 57:241-258. 60 Merila, J., and B. C. Sheldon. 2000. Lifetime reproductive succes and heritability in nature. American Naturalist 155:301-310. Michener, G. R. 1989. Reproductive efort during gestation and lactation by Richardson's ground squirels. Oecologia 78:77-86. Milesi, E., S. Huber, L. Everts, and J. Ditami. 1999. Reproductive decisions in female European ground squirels: Factors afecting reproductive output and maternal investment Ethology 105:163-175 Moore, R. W., E. J. Eisen, and L. C. Ulberg. 1970. Prenatal and postnatal maternal influences on growth in mice selected for body weight. Genetics 64:59-68. Mousseau, T. A., and H. Dingle. 1991. Maternal efects in insect life histories Annual Review of Entomology 36:511-534. Mousseau, T. A., and C. W. Fox. 1998. The adaptive significance of maternal efects. Trends in Ecology and Evolution 13:403-407. Murie, J. O. 1992. Predation by badgers on Columbian ground squirels. Journal of Mamalogy 73:385-394. Murie, J. O. 1995. Mating behavior of Columbian ground squirels. I. Multiple mating by females and multiple paternity Canadian Journal of Zoology 73:1819-1826. Murie, J. O., and D. A. Boag. 1984. The relationship of body weight to overwinter survival in Columbian ground squirels. Journal of Mamalogy 65:688-690. Murie, J. O., and F. S. Dobson. 1987. The costs of reproduction in female columbian ground squirels. Oecologia 73:1-6. 61 Murie, J. O., and M. A. Haris. 1982. Annual variation of spring emergence and breding in Columbian ground squirels (Spermophilus columbianus). Journal of Mamalogy 63:431-439. Murie, J. O., and M. A. Haris. 1988. Social interactions and dominance relationships betwen female and male Columbian ground squirels. Canadian Journal of Zoology 66:1414-1420. Murie, J. O., S. D. Stevens, and B. Leoppky. 1998. Survival of captive-born cross- fostered juvenile Columbian ground squirels in the field. Journal of Mamalogy 79:1152-1160. Myers, P., and L. L. Master. 1983. Reproduction by Peromyscus maniculatus: size and compromise. Journal of Mamalogy 64:1-18. Myers, R. H. 1990. Clasical and modern regresion with applications, 2 edition. PWS Kent, Boston. Nager, R. G., P. Monaghan, R. Grifiths, D. C. Houston, and R. Dawson. 1999. Experimental demonstration that ofspring sex ratio varies with maternal condition Procedings of the National Academy of Sciences 96:570-573. Parichy, D. M., and R. H. Kaplan. 1992. Maternal efects on offspring growth and development depend on environmental quality in the frog Bombina orientalis. Oecologia 91:579-586. Pedhazur, E. J. 1982. Multiple regresion in behavioral research. Holt, Rinehart, and Winston, New York, New York. 62 Pelayo, J. T., and R. G. Clark. 2003. Consequences of egg size for offspring survival: a cross-fostering experiment in ruddy ducks (Oxyura jamaicensis). Auk 120:384- 393. Perins, C. M., and R. H. McClery. 2001. The efect of fledging mas on the lives of Great tits Parus major. Ardea 89:135-142. Petraitis, P. S., A. E. Dunham, and P. H. Niewiarowski. 1996. Infering multiple causality: the limitations of path analysis and multiple regresions. Functional ecology 10:421-431. Price, T. 1998. Maternal and paternal efects in birds. Pages 202-226 in T. A. Mousseau and C. W. Fox, editors. Maternal efects as adaptations. Oxford University Pres, New York. Rhees, B. K., C. A. Ernst, C. H. Miao, and W. R. Atchley. 1999. Uterine and postnatal maternal efects in mice selected for diferential rate of early development. Genetics 153:905-917. Rieger, J. F. 1996. Body size, liter size, timing of reproduction, and juvenile survival in the Uinta ground squirel, Spermophilus armatus. Oecologia 107:463-468. Risch, T. S., F. S. Dobson, and J. O. Murie. 1995. Is mean liter size the most productive? A test in Columbian ground squirels. Ecology 76:1643-1654. Riska, B., W. R. Atchley, and J. J. Rutledge. 1984. A genetic analysis of targeted growth in mice. Genetics 107:79-101. Roach, D. A., and R. D. Wulf. 1987. Maternal efects in plants. Annual Review of Ecology and Systematics 18:209-235. 63 Roff, D. A. 1998. The detection and measurement of maternal efects. Pages 83-96 in T. A. Mousseau and C. W. Fox, editors. Maternal efects as adaptations. Oxford University Pres, New York. Rutledge, J. J., R. O.W., E. J. Eisen, and J. E. Legates. 1972. Dynamics of genetic and maternal efects in mice. Journal of Animal Science 35:911-918. SAS Institute, I. 2002. SAS v.9.1 for Windows. in, Cary, N.C. Schulte-Hostedde, A. I., B. Zinner, J. S. Milar, and G. J. Hickling. 2005. Restitution of mas-size residuals: validating body condition indices. Ecology 86:155-163. Shaw, R. G., and D. Byers, L. 1998. Genetics of maternal and paternal efects. Pages 97- 111 in T. A. Mousseau and C. W. Fox, editors. Maternal efects as adaptations. Oxford University Pres, New York. Skibiel, A. L. 2007. Maternal efects in Columbian ground squirels. Master's thesis. Auburn University Skogland, T. 1984. The efects of food and maternal conditions on fetal growth and size in wild reindeer. Rangifer 4:39-46. Smith, H. G., and K.-J. Wetermark. 1995. Heritability of nestling growth in cross- fostered European starlings Sturnus vulgaris. Genetics 141:657-665. Sokal, R. R., and F. J. Rohlf. 1995. Biometry, 3rd edition edition. W. H. Freman and Company, New York. Stearns, S. C. 1992. The evolution of life histories. Oxford University Pres, New York. Trivers, R. L., and D. E. Wilard. 1973. Natural selection of parental ability to vary sex ratio of offspring. Science 179:90-92. 64 Wilson, D. E., and S. Ruff, editors. 1999. The Smithsonian Book of North American Mamals. Smithsonian Institution Pres, Washington. Wolf, J. B., E. D. I. Brodie, J. M. Cheverud, A. J. Moore, and M. J. Wade. 1998. Evolutionary consequences of indirect genetic efects. Trends in Ecology and Evolution 13:64-69. Zwank, P. J., and J. A. Zeno. 1986. Weight gain of white-tailed deer fawns relative to fawning date. Procedings of the Annual Conference of the Southeastern Asociation of Fish and Wildlife Agencies 40:424-429.