Oecologia (1999) 121:19±24
Ó Springer-Verlag 1999
Susanne Huber á Eva Millesi á Manfred Walzl John Dittami á Walter Arnold
Reproductive effort and costs of reproduction in female European ground squirrels
Received: 22 March 1999 / Accepted: 7 June 1999
Abstract Reproductive eort, factors aecting reproductive output and costs of reproduction were studied in primiparous yearling compared to multiparous older female European ground squirrels (Spermophilus citellus). Yearling females weaned smaller litters than older ones. Litter size increased with posthibernation body mass at the expense of slightly lighter young for yearling but not for older mothers. In older females, on the other hand, emergence body mass in¯uenced ospring mass, whereas litter size was aected by oestrus date. High reproductive eort entailed reproductive costs in terms of reduced subsequent fecundity but not subsequent survival for both yearling and older females. The production of large litters and long duration of lactation delayed subsequent oestrus, which, in turn, correlated negatively with litter size. During the second half of lactation, oestradiol levels were signi®cantly elevated, indicating the initiation of follicular maturation processes. Oestradiol levels during that time correlated negatively with current, but positively with subsequent litter size. We therefore assume that inhibitory eects of lactation on gonadal development may mediate the negative relationship between reproductive eort and subsequent reproductive timing in adults. This eect is absent in yearlings because they are reproducing for the ®rst time. Reproductive output in yearlings was in¯uenced by interactions between structural growth and puberty. Key words Ground squirrel á Female á Reproduction á Lactation á Gonad development S. Huber (&) Research Institute of Wildlife Ecology, University of Veterinary Medicine of Vienna, Savoyenstrasse 1, A-1160 Wien, Austria e-mail:
[email protected] E. Millesi á M. Walzl á J. Dittami Institute of Zoology, University of Vienna, Althanstrasse 14, A-1090 Wien, Austria
Introduction Life history theory predicts a ``trade-o'' between current reproductive eort and subsequent reproduction and/or survival. This trade-o has been termed the costof-reproduction hypothesis (Lack 1966; Williams 1966; Stearns 1976; Bell 1980; Partridge and Harvey 1985). Costs of reproduction are separated into survival costs, where current reproduction decreases an individual's probability of future survival, and fecundity costs, where current reproductive eort aects future reproductive capacity (Reznick 1992). The mechanisms by which reproductive eort is carried over into subsequent reproductive output, however, often remain unclear. Although most studies in this ®eld have been conducted in birds, recently reproductive costs have also been analysed in mammals (Mappes et al. 1995; Koskela 1998; Koskela et al. 1998). In European ground squirrels (Spermophilus citellus), reproduction is constrained by hibernation, limiting both the timing of mating and gonadal development. Females produce one litter per year and mate within a few days after emergence from hibernation in order to maximise the time available for their ospring to grow and fatten prior to hibernation (Millesi et al. 1999a). Gonadal development is limited by hibernation as maturation processes are suppressed during hypothermia (Barnes et al. 1986). Follicular maturation, however, is known to be a long-term process (reviewed in Greenwald and Roy 1994). This leads to the assumption that in European ground squirrels, follicular development may be initiated during summer prior to the hibernation period. As a consequence, interactions with other activities in one season would aect these maturation processes and reproductive output in the next season. Lactation, for example, is known to depress luteinising hormone (LH) secretion, thereby suppressing the LH-dependent processes of gonadal development (reviewed in Mc Neilly 1994). Hence, physiological constraints limiting maturation processes possibly me-
20
diate costs of reproduction. Female European ground squirrels usually breed as yearlings (Millesi et al. 1999a), but ®rst breeders often produce smaller litters than older ones (e.g. Festa-Bianchet and King 1991; Dobson and Michener 1995). While yearlings breed for the ®rst time, older females have already reproduced during the previous year. Consequently, functional aspects of reproductive decisions may dier between primiparous yearling and multiparous older individuals. Our study was aimed (1) at examining reproductive eort ± measured in terms of litter size, weanling mass and the duration of lactation ± and factors aecting reproductive output in primiparous yearling compared to multiparous older females and (2) at investigating costs of reproductive eort in terms of reduced survival and/or fecundity in both age groups.
Materials and methods The study was carried out in a recreation park near Vienna in 1996 and 1997 and was based on long-term research in this population (Millesi et al. 1998, 1999a, b). According to Millesi et al. (1999a), almost all females of the focal population reproduced (94%) without dierences between age classes. Consequently, all older females were multiparous. We monitored the focal population by daily scans supplemented with all additional sightings not recorded during scans (Millesi et al. 1999a). We trapped the animals in Tomahawk live traps baited with peanut butter. Each female (1996: n = 29; 1997: n = 25) was repeatedly trapped in 3- to 5-day-intervals for 3 weeks after emergence from hibernation in April to determine oestrus date, and at 2- to 4-week intervals thereafter until the beginning of hibernation in August with a focus on reproductive phases (gestation, ®rst and second half of lactation, postlactation) and on critical events like parturition, weaning and onset of prehibernatory fattening. Unfortunately, we did not succeed capturing all individual females at each stage. Therefore, sample sizes vary for the dierent parameters. After ®rst capture, we subcutaneously implanted in the squirrels a transponder chip (PIT tag, Indexel) for permanent identi®cation. For distant recognition, the animals were marked with a commercial hair dye in individually recognisable patterns. Body mass was measured with a laboratory scale, and scull size (maximal length of scull) was determined by a vernier calliper. We de®ned spring emergence body mass as the body mass measured within 5 days after emergence from hibernation. Reproductive state was determined by assessing teat development, milk excretion, and rapid body mass changes as indicators of littering (Millesi et al. 1999a). At capture, a 100-ll blood sample was taken from the femural vein for oestrogen analysis with a minimum interval of 2 weeks between two successive blood collections. As follicular maturation is accompanied by elevated oestrogen secretion (Ojeda 1996), oestrogen levels can indicate follicular maturation processes. The blood samples were centrifugated in the ®eld. Plasma was then taken into the laboratory and stored at )20°C. Oestradiol levels were analysed after diethyl ether extraction with a biotin-streptavidin enzyme immunoassay according to Palme and MoÈstl (1993). Intra- and interassay variation were less than 10%. We avoided any adverse eects on the animals by our procedures. Animals were taken out of the traps immediately after capture and put into a black bag, where they were always calm. To avoid stressing the squirrels, we took only the manipulated parts of the body out of the bag. Together with determination of body mass and reproductive state, all manipulation procedures lasted less than 5 min. The animals were released immediately afterwards at the trapping location and resumed their former behaviour either at once, or after they have hidden in a burrow for a few minutes. We
examined the eects of repeated trapping and the manipulations on body mass, behaviour and reproductive output to rule out capturing eects on the results. By comparing these parameters in focal individuals and females that had been captured rarely without further manipulations, we did not ®nd any dierences between the two groups. The animals were categorised into three age classes: juvenile (prior to ®rst hibernation), yearling (after ®rst hibernation) and older females (at least 2 years old). This was possible since we know the year of birth of most of the focal individuals. For a few females for which we did not know the exact year of birth, we categorised yearling and older females according to scull length, as yearlings have signi®cantly smaller sculls than older females (Millesi et al. 1999b). Survival of individual females to the next breeding season was determined at emergence from hibernation. We could, however, not dierentiate between mortality and dispersal in each case, although in many cases we observed kills by domestic cats. We determined time of vaginal oestrous from vaginal smears collected during the ®rst 3 weeks after emergence from hibernation, following Holmes and Landau (1986). Smears were stained using the Papanicolaou stain protocol and cytology was classi®ed under a Polyvar microscope by the ratios and numbers of nucleated epithelial cells, corni®ed cells and leucocytes. Lactation was de®ned as the period from parturition until the onset of teat regression (when teats are still large but limp and dried out). Estimates of the end of lactation were supported by observations of burrow use. Females stopped using litter burrows after weaning of young. Litter size was determined when juveniles emerged from natal burrows for the ®rst time, which never coincided with weaning. As individual burrows were only used by a single female, litter size was de®ned as the number of juveniles emerging from a female's burrow. Litter mass was calculated as litter size times mean juvenile emergence body mass. X-ray analyses on intrauterine litter size showed that the number of emerging ospring is determined prenatally and that dierential postnatal mortality is unlikely to explain litter size variation (Millesi et al. 1999a). We therefore used emergence litter size as a measure of fecundity. Statistical analysis was conducted using the SPSS 7.5 for Windows statistical package. In 1996, the onset of reproduction was delayed due to 10 days of heavy snowfall. To provide comparability of data, we corrected for this delay. If assumptions were met, we applied parametric tests. For some analyses this was the case only after square root transformation of the original data. We used individual means when repeated measurements were available in a particular phase. Linear regressions and ANCOVA for regression were used to compare reproductive eort and fertility costs between yearling and older females. Logistic regression was performed to test survival costs of reproduction. Signi®cance values were obtained from two-tailed statistics. Means are presented 1SD.
Results Reproductive output Yearling females weaned signi®cantly smaller litters (5.36 1.86 young, range 3±8, n = 14) than older mothers (6.8 1.72 young, range 3±10, n = 26; t-test for independent samples: t38 = )2.47, P = 0.018). Mean oestrus date was later in yearling (April 12 1.8, n = 14) than older females (April 7 4.7, n = 26; t-test for independent samples: t38 = 3.474, P = 0.001). Body mass at emergence from hibernation also diered signi®cantly between yearling (175.1 27.42 g, n = 12) and older females (199.5 20.19 g, n = 18; t-test for independent samples: t28 = )2.813, P = 0.009).
21
In yearling females, emergence body mass correlated positively with litter size (Fig. 1), suggesting a bodymass-dependent adjustment of reproductive output. In older females, however, this association was not found (Fig. 1). An opposite picture emerged from analysing the relation between litter size and oestrus date. While litter size decreased with progressing oestrus date in older females (Fig. 2), this eect seemed to be absent in yearlings (Fig. 2), though the regression slopes did not dier signi®cantly. Yearling and older mothers allocated reproductive eort dierently into number and weight of ospring.
Heavier yearling mothers invested in additional but lighter young (cf. Fig. 1 and Fig. 3). In contrast, heavier older females did not invest in more, but in heavier young (cf. Fig. 1 and Fig. 3). Accordingly, female posthibernation body mass showed a signi®cant positive correlation with mean ospring body mass at natal emergence after correction for litter size for older but not for yearling females (partial correlation coecient: mean juvenile body mass at natal emergence versus female emergence body mass, corrected for litter size: older females, r = 0.5323, n = 15, P = 0.05; yearling females, r = 0.3845, n = 7, P = 0.452).
Fig. 1 Litter size versus body mass at emergence from hibernation (g) in older females (closed circles), and yearling females (open squares). Regression lines: older females, y = 7.842 ) 0.0056x, R2 = 0.004, t19 = )0.286, P = 0.78; yearling females: y = )2.671 + 0.0474x, R2 = 0.584, t8 = 3.351, P = 0.01. Slopes of regression lines are signi®cantly dierent between yearling and older females (ANCOVA: t27 = 2.083, P = 0.047)
Fig. 3 Mean individual body mass in litters at natal emergence (g) versus female body mass at emergence from hibernation (g) for older females (closed circles), and yearling females (open squares). Regression lines: older females, y = )34.833 + 0.491x, R2 = 0.301, t13 = 2.364, P = 0.034; yearling females, y = 121.63 ) 0.295x, R2 = 0.30, t5 = )1.491, P = 0.196. Slopes of regression lines are signi®cantly dierent between yearling and older females (ANCOVA: t18 = 2.793, P = 0.012)
Fig. 2 Litter size versus oestrus date in older females (closed circles), and yearling females (open squares). Regression lines: older females, y = 24.602 ) 0.17x, R2 = 0.187, t24 = )2.352, P = 0.027; yearling females: y = 9.43 ) 0.04x, R2 = 0.002, t12 = )0.169, P = 0.87. Slopes of regression lines are not signi®cantly dierent between yearling and older females (ANCOVA: t36 = 0.817, P = 0.419)
Fig. 4 Oestrus date in 1997 versus litter size in 1996 in older (closed circles) and yearling (open squares) females. Regression lines: older females, y = 90.35 + 2.77x, R2 = 0.598, t5 = 2.729, P = 0.041; yearling females, y = 99.9 + 0.779x, R2 = 0.599, t5 = 2.731, P = 0.041. Slopes of regression lines show a tendency to be dierent between yearling and older females (ANCOVA: t10 = 2.082, P = 0.064)
22
Costs of reproduction Producing large litters delayed oestrus in the following year in both yearling and older females (Fig. 4). A similar relationship was found between litter mass and subsequent oestrus date (linear regression: R2 = 0.557, t7 = 2.969, P = 0.021). We pooled yearling and older females for this and the following analyses because age classes did not dier signi®cantly in either case. Duration of lactation in one season aected oestrus delay in the next. Females nursed large litters signi®cantly longer than smaller ones (Pearson correlation: r = 0.798, n = 22, P < 0.001). Long-lactating females, in turn, reproduced later in the subsequent season, i.e. weaning date was positively correlated with the following oestrus
date (Fig. 5). Lactation duration correlated positively with subsequent oestrus date as well (Pearson correlation: r = 0.605, n = 14, P = 0.022). During the second half of lactation, mean oestradiol levels were signi®cantly elevated (Fig. 6). Females that nursed smaller litters had higher oestrogen levels during that time than did mothers lactating larger litters (Fig. 7a). On the other hand, subsequent litter size correlated positively with oestrogen levels during the second half of lactation. Females with high oestrogen titres produced larger litters the following year (Fig. 7b). We did not ®nd any eect of reproductive eort on survival. A mother's probability of surviving to the subsequent breeding period was neither signi®cantly in¯uenced by litter size (logistic regression: b = )0.078, SE = 0.177, n = 40, P = 0.66) nor by mean young emergence body mass (logistic regression: b = 0.0045, SE = 0.023, n = 26, P = 0.84), litter mass (logistic regression: b = 0.0008, SE = 0.0052, n = 30, P = 0.88) or lactation duration (logistic regression: b = )0.0009, SE = 0.1404, n = 22, P = 0.99). As age classes did not dier signi®cantly, they were not separated for analysis.
Discussion Although yearling female European ground squirrels reproduced at a similar rate as older females, reproductive performance diered markedly between the two age groups. Yearling females not only produced smaller litters compared to older ones, their litter size correlated Fig. 5 Oestrus date in 1997 versus weaning date in 1996 in older with posthibernation body mass whereas it did not in females (closed circles), and yearling females (open squares). Regression line (yearling plus older females): y = 50.66 + 0.248x, R2 = older mothers. Presumably as a result of this correlation, 0.888, t6 = 6.907, P < 0.001 yearling posthibernation body mass had a slightly negative eect on neonate mass. Hence, heavy yearlings invested in additional ospring at the expense of slightly
Fig. 6 Oestradiol titres (ng/ml) during gestation (G), ®rst half of lactation (L1, de®ned as the period from the ®rst to the medium day of lactation in each individual), second half of lactation (L2, de®ned as the period from the medium to the last day of lactation in each individual), and postlactation (PL). Means + 1 SD are shown. Phase dierences (data were square root transformed for analysis to correct for non-normality): ANOVA, F3,64 = 2.542, P = 0.064; post-hoc comparisons (LSD), L2±G: P = 0.032, L2±L1: P = 0.044, L2±PL: P = 0.011; all others: P > 0.5. Sample sizes: G = 11, L1 = 20, L2 = 12, PL = 25
Fig. 7 a Oestradiol titres during the second half of lactation in 1996 (ng/ml) versus litter size in 1996 in older (closed circles) and yearling (open squares) females. Regression line (yearling plus older females): y = 0.113 ) 0.011x, R2 = 0.242, t11 = )1.873, P = 0.088. b Litter size in 1997 versus oestradiol titres during the second half of lactation in 1996 (ng/ml) in older (closed circles) and yearling (open squares) females. Regression line (yearling plus older females): y = )0.03 + 0.0156x, R2 = 0.476, t6 = 2.337, P = 0.058
23
smaller young. The relationship between body mass and litter size in yearling females may be due to the fact that yearlings enter hibernation as juveniles before reaching maturity. So before breeding they have to pass through puberty which is known to in¯uence and be in¯uenced by structural growth (reviewed in Foster 1994): additional development or improved nutrition are translated to the gonadotropin-releasing hormone-pulse generator to increase its activity. The resultant increase in LH pulse frequency drives the pubertal processes by developing follicles (reviewed in Foster 1994). Puberty also seems to be the main reason for the later mean oestrus date in yearling compared to older females. In older females, litter size did not vary with body mass but with oestrus date. This negative relation between litter size and oestrus date possibly represents carryover eects from previous parental eort to subsequent reproduction. We suggest that physiological eects of previous lactation in¯uenced subsequent oestrus date by aecting follicular maturation processes (see below). This is supported by the fact that, in contrast to yearlings, older females with high posthibernation body mass produced heavier instead of additional young. Reproductive output diers between yearling and older females in a number of ground squirrel species (e.g. Armitage and Downhower 1974; Sherman and Morton 1984; Festa-Bianchet and King 1991; Dobson and Michener 1995). In Richardson's ground squirrels, a similar pattern as seen in our study has been described: litter size and maternal body mass correlated among yearlings, but not among older females, while litter size decreased signi®cantly as the breeding season progressed in older, though not yearling females. Moreover, in yearling females, body mass in¯uenced litter size, while in older females, body mass aected neonate mass (Michener 1989; Dobson and Michener 1995). Dobson and Michener (1995) also showed that in yearling females, dierences in body mass primarily re¯ected dierences in structural size, whereas body mass of older females was strongly associated with condition, i.e. fat reserves. If a similar principle applies to the European ground squirrel ± which remains to be examined ± this would support our suggestions that structural growth in yearling females aected reproductive output through an eect on litter size, while body condition in older ones had a minor eect on litter size but was re¯ected in ospring mass. Ospring body mass, in turn, may aect juvenile survival to the following spring (e.g. Murie and Boag 1984; Michener and Locklear 1990) and/or subsequent juvenile reproductive success (e.g. Michener 1989; King et al. 1991; Rieger 1996). In our study, ospring mass had no eect on juvenile survival but in¯uenced juvenile reproductive output in the following season (S. Huber, I.E. Homann, E. Millesi, J. Dittami, W. Arnold, unpublished work). Thus, yearling females ± allocating surplus resources to the production of additional but lighter young ± had ®tness costs in terms of reduced ospring fecundity, while older females ± allocating surplus re-
sources to the production of heavier young ± had ®tness bene®ts in terms of increased reproductive output of their ospring. Both yearling and older females weaning large litters had reproductive costs in terms of reduced subsequent fecundity. The subsequent oestrus of females producing large and heavy litters was delayed. Subsequent oestrus date, in turn, was negatively correlated with litter size in multiparous females. Direct eects of current reproductive output on future litter size were less evident in our study. During a former study on the same population (J. Dittami, E. Millesi, S. Huber, E. MoÈstl, M. Walzl, unpublished data), however, a signi®cant inverse relationship between sequential litter sizes was found. Hence, following Trivers (1972), high parental investment in terms of long-lactating large litters reduced the residual reproductive value, representing a ``trade-o'' between current and subsequent reproduction. On the other hand, females did not pay reproductive costs in terms of reduced survival. Females weaning large or heavy litters were as likely to survive to the subsequent mating period as mothers of smaller litters and young. Correspondingly, in Richardson's and Columbian ground squirrels, large litters did not impose higher survival costs on the mother than smaller ones (Murie and Dobson 1987; Michener and Locklear 1990; Risch et al. 1995). In contrast to our ®ndings, however, ground squirrel studies to date have not described fecundity costs, as no eects of successful reproduction on future reproductive performance were found (Michener and Locklear 1990; Hare and Murie 1992; Risch et al. 1995). In other small mammals like the bank vole (Clethrionomys glareolus) also, high reproductive eort caused by litter size manipulation had no eect on future maternal survival or reproduction (Mappes et al. 1995; Koskela 1998). The relationship between reproductive eort and timing of subsequent oestrus in our study may be mediated by the reproductive physiology of these animals. Large litters were nursed longer than smaller ones, and lactation in one season aected oestrus delay in the next. Both late weaning date and long duration of lactation postponed oestrus date in the subsequent season. During late lactation, oestradiol levels were signi®cantly elevated, indicating initiation of follicular maturation processes during that time. Preliminary histological data support this suggestion, as Graa®an follicles were present in late summer ovaries (J. Dittami, E. Millesi, S. Huber, E. MoÈstl, M. Walzl, unpublished data). In addition, high oestradiol levels during the second half of lactation were negatively related to current, but positively related to subsequent litter size. We assume that follicular development may be impaired due to a higher investment in large litters by long (and probably more intense) lactation, which would lead to an oestrus delay in the following mating period. Follicular development is a longterm process characterised by changes in follicle size and morphology as well as by ¯uctuations in sex hormone levels (reviewed in Greenwald and Roy 1994). Interac-
24
tions between follicular maturation and other activities can have profound eects on developmental processes. LH secretion, for example, is highly depressed during lactation, suppressing the LH-dependent processes of follicular development (reviewed in Mc Neilly 1994). This LH de®ciency decreases steroid production by the follicle (reviewed in Greenwald and Roy 1994). In animals kept in the laboratory, suppression of follicular maturation during lactation has been clearly demonstrated. Moreover, the dierences in suckling stimulus between small and large litters in¯uenced the degree of follicular inhibition in the rat. This inhibition was also re¯ected in low levels of oestrogen (reviewed in Mc Neilly 1994). Thus, it seems reasonable to assume that in European ground squirrels, follicular maturation is impaired by long and intense lactation, resulting in low oestrogen levels and a delay in subsequent reproduction. Acknowledgments This work was supported by the Austrian National Bank (JubilaÈumsfonds 6264). We thank I.E. Homann and S. Hanslick for their help in collecting ®eld data as well as the referees for their helpful comments on the manuscript. We declare that the experiments comply with the current laws of Austria.
References Armitage KB, Downhower JF (1974) Demography of yellow-bellied marmot populations. Ecology 55:1233±1245 Barnes BM, Kretzmann M, Licht P, Zucker I (1986) The in¯uence of hibernation on testis growth and spermatogenesis in the golden-mantled ground squirrel, Spermophilus lateralis. Biol Reprod 35:1289±1297 Bell G (1980) The costs of reproduction and their consequences. Am Nat 116:45±76 Dobson FS, Michener GR (1995) Maternal traits and reproduction in Richardson's ground squirrels. Ecology 76:851±862 Festa-Bianchet M, King WJ (1991) Eects of litter size and population dynamics on juvenile and maternal survival in Columbian ground squirrels. J Anim Ecol 60:1077±1090 Foster DL (1994) Puberty in the sheep. In: Knobil E, Neill JD (eds) The physiology of reproduction. II. Raven, New York, pp 411± 453 Greenwald GS, Roy SK (1994) Follicular development and its control. In: Knobil E, Neill JD (eds) The physiology of reproduction. I. Raven, New York, pp 629±725 Hare JF, Murie JO (1992) Manipulation of litter size reveals no cost of reproduction in Columbian ground squirrels. J Mammal 73:449±454 Holmes WG, Landau IT (1986) Vaginal estrus in unmated Belding's ground squirrels. Horm Behav 20:243±248 King WJ, Festa-Bianchet M, Hat®eld SE (1991) Determinants of reproductive success in female Columbian ground squirrels. Oecologia 86:528±534 Koskela E (1998) Ospring growth, survival and reproductive success in the bank vole: a litter size manipulation experiment. Oecologia 115:379±384
Koskela E, Jonsson P, Hartikainen T, Mappes T (1998) Limitation of reproductive success by food availability and litter size in the bank vole, Clethrionomys glareolus. Proc R Soc Lond B 264:1129±1134 Lack D (1966) Population studies of birds. Clarendon, Oxford Mappes T, Koskela E, YloÈnen H (1995) Reproductive costs and litter size in the bank vole. Proc R Soc Lond B 261:19±24 Mc Neilly AS (1994) Suckling and the control of gonadotropin secretion In: Knobil E, Neill JD (eds) The physiology of reproduction. II. Raven, New York, pp 1179±1213 Michener GR (1989) Reproductive eort during gestation and lactation by Richardson's ground squirrels. Oecologia 78:77±86 Michener GR, Locklear L (1990) Dierential costs of reproductive eort for male and female Richardson's ground squirrels. Ecology 71:855±868 Millesi E, Huber S, Everts LG, Dittami JP (1999a) Reproductive decisions in female European ground squirrels: factors aecting reproductive output and maternal investment. Ethology 105:163±175 Millesi E, Strijkstra AM, Homann IE, Dittami JP, Daan S (1999b) Sex and age dierences in mass, morphology and annual cycle in European ground squirrels, Spermophilus citellus. J Mammal 80:218±231 Millesi E, Huber S, Dittami JP, Homann IE, Daan S (1998) Parameters of mating eort and success in male European ground squirrels, Spermophilus citellus. Ethology 104:298±313 Murie JO, Boag DA (1984) The relationship of body weight to overwinter survival in Columbian ground squirrels. J Mammal 65:688±690 Murie JO, Dobson FS (1987) The costs of reproduction in female Columbian ground squirrels Spermophilus columbianus. Oecologia 73:1±6 Ojeda SR (1996) Female reproductive function. In: Grin JE, Ojeda SR (eds) Textbook of endocrine physiology. Oxford University Press, New York, pp 164±200 Palme R, MoÈstl E (1993) Biotin-streptavidin enzyme immunoassay for the determination of oestrogens and androgens in boar faeces. Proc 5th Symp Analysis Steroids, pp 111±117 Partridge L, Harvey PH (1985) Costs of reproduction. Nature 316:20 Reznick D (1992) Measuring the costs of reproduction. Trends Ecol Evol 7:42±45 Rieger JF (1996) Body size, litter size, timing of reproduction, and juvenile survival in the Uinta ground squirrel, Spermophilus armatus. Oecologia 107:463±468 Risch TS, Dobson FS, Murie JO (1995) Is mean litter size the most productive? A test in Columbian ground squirrels. Ecology 76:1643±1654 Sherman PW, Morton ML (1984) Demography of Belding's ground squirrels. Ecology 65:1617±1628 Stearns S (1976) Life-history tactics: a review of ideas. Q Rev Biol 51:3±47 Trivers RL (1972) Parental investment and sexual selection. In: Campbell B (ed) Sexual selection and the descent of man. Heinemann, London, pp 136±179 Williams GC (1966) Natural selection, the cost of reproduction and a re®nement of Lack's principle. Am Nat 100:687±690
