There is great variation in the age at which females of different mammalian species first breed. Recent comparative analyses have focused on the relationship between age at first reproduction and body size, but differences in patterns of mortality experienced by natural populations are expected to have major effects on selection for age at first reproduction. Here we show that the age at which females first reproduce is strongly correlated with expectation of life at birth, after the effects of body size have been removed, within and among species of mammals living in natural populations.

We quantitatively test theoretical predictions concerning mammalian life histories, using published data on survival, reproduction, and body mass for 29 eutherian mammals. Larger mammals have a greater age at maturity, greater generation length, greater life expectancy, lower reproductive value at maturity, and smaller litters than do smaller mammals. Residual reproductive value at maturity is not correlated with adult body mass or survival. Litter size varies inversely with generation length and adult survival. Age at maturity is positively correlated with life expectancy. Twenty-seven of 29 mammals display a generation length longer than their life expectancy at birth, and the same proportion shows a greater life expectancy at maturity than at birth. A fairly high proportion (76-82%) of the variation in these dependent variables is attributable to adult mass.Many life table characteristics of mammals are interrelated, although not necessarily in the ways predicted by theory. Design constraints may preclude significant differences in life history patterns among mammals, so that the life table characteristics of only a few species may depict the pattern of life table evolution in most eutherian mammals.

We quantitatively tested various aspects of the theory of r— and K—selection for six populations of Columbian ground squirrels from Alberta, Canada. Three measures of environmental predictability (maximum and minimum temperatures, precipitation) and a presumed measure of food resource levels supported the prediction that environments at lower elevations were less predictable, and had greater, more variable food resource levels than environments at higher elevations. Columbian ground squirrels in more predictable environments (i.e., at higher elevations) had higher adult survival rates, later ages at maturity, and possibly lower genetic variabilities than did squirrels in less predictable environments (at lower elevations). Body mass was greater at lower elevations than at higher elevations. Litter size showed no trend with respect to elevation, but it tended to be most variable in unpredictable environments at lower elevations. Although they were more predictable, the daily minimum temperatures at higher elevations tended (P = .06) to show wider variation than the minimum temperatures at lower elevations. Previous apparent problems with r—K theory may be attributable in part to the assumption that predictability and stability should covary. The major difference between the pattern emerging from our study and that predicted by traditional theory is that predictability of environments, and concomitantly the occurrence of K—strategists, was found to increase with movement up an elevational gradient. Portions of r—K theory may be found to be useful once all of its parameters are consistently measured.

Eleven theoretical predictions (or assumptions) oflife history evolution are considered for the montane Columbian ground squirrel, Spermophilus columbianus, using age-specific survival and fecundity from six life tables of natural populations. The following statements are supported among age classes among populations, among age classes within populations, and (or) within age classes among populations: (1) mortality rates are high a~ter birth, drop to a minimum by age 1 yr, and then rise with age; (2) fecundity increases wtth age and seldom decreases at the last age of reproduction; (3) reproductive value and residual reproductive value rise to a peak and then fall with age; (4) age-specific mortality rates and age-specific mortality covary inversely with reproductive value; (5) residual reproductive value, survival, and survival rates covary inversely with fecundity; (6) residual reproductive value is positively correlated with adult survival; (7) no relationships were found between fecundity and successive survival probabilities in the life table; (8) no relationship was found between age at maturity and life expectancy; (9) no relationship was found between litter size and generation length; ( 1 0) future fecundity is positively correlated with present fecundity; and (11) age-specific fecundity varies inversely with modified reproductive value. Life history patterns among populations, within populations within age classes, and among species are not always similar, so that theoretical prediction~ should explicitly delineate the level of organization to which they pertain.

Eight years of age-specific survival data and 6 years of fecundity data from a free-living population of Belding's ground squirrels (Spermophilus beldingi) at Tioga Pass, California, were used to test the hypothesis that time-specific life tables, based on data from individual years, were different from the cohort-specific life table, based on the combined data from all years. The results indicated that neither the age structure of the male nor the female population significantly differed among years (all P > 0.05). Furthermore, the means and the variances in the sizes of weaned litters did not differ among years either in the population at large or within individual age-classes (all P > 0.05). A 27-day snowstorm that occurred in the spring of 1977 increased mortality and reduced reproduction, but it did not change the ground squirrels' age-specific survival or fecundity patterns. Taken together, our analyses revealed that each time-specific life table provided age-specific survival and fecundity estimates that were statistically indistinguishable (P > 0.05) from the composite, cohort-specific life table for each sex, regardless of severe environmental conditions. This is the first demonstration of the equivalence of time- and cohort-specific life tables for a free-living population of mammals.