## Life Tables

Life tables were invented by the life insurance industry as a means of estimating riskāthat is, how long a certain person can be expected to pay the premiums on a life insurance policy before making the claim. Animal ecologists took over the idea when they saw how useful it was as a means of making standardized comparisons within and among populations. A life table is a useful way to estimate the variation in survival rates (and, conversely, the mortality rates) among the age classes in a population. Many excellent books explain how to construct and analyze the tables, if we can get that far. The trouble with weasels is that few people have been able to get over the difficulties of collecting the valid samples needed for constructing the tables in the first place.

The easiest way to construct a life table requires that the rate of increase of the population is zero (density is not changing) and that the age structure is stable (constant proportions of ages in every generation). When these two conditions are met, a population is said to be stationary, and the standard formulae can be applied. Neither condition is ever met by weasel populations. This does not mean that life tables cannot be constructed for weasels, only that they are harder to calculate and must be interpreted cautiously.

A frequency distribution of age classes for a stationary population is proportional to the probability of living to each age. Two methods of sampling wild populations can be used to collect the age distribution data. Both methods can be used to estimate the probability that a newborn will live to a given age, but different assumptions lie behind each of the two methods.

The first is to catch a group of newborn young representing a particular cohort, mark and release them, and then watch them from a distance to see at what age each dies from natural causes. At some times of year it is not difficult to catch young weasels in live traps and to mark them with eartags. Keeping track of them afterward, however, is very difficult, especially the young males, which may disperse over great distances in their first year (Chapter 9).

The other way is to collect a large sample of dead weasels, taking care to make the traps equally available to both sexes and all ages, and then work out the age of each on the day it was killed. Catching weasels in steel kill traps is usually not difficult, but making the traps equally available to both sexes and all ages is nearly impossible, and estimating weasels' ages has been, until recently, easier said than done, especially in the adults.

Of course, both methods have problems. For example, if kill trapping is used, the complete age structure can be obtained from one or a few samples, but if removal sampling affects the longevity of the older adults in the target population, data from a previously untrapped population may be slightly different from data from regularly trapped populations. The same caution does not apply to the proportion of the young of the year in a sample, because it is controlled almost entirely by the success of the previous breeding season, not by trapping history (Figure 11.3).

Conversely, if live trapping is used, the sampling method has minimum effect on natural age structure, but the ages of the oldest adults can be known only if the study lasts long enough to follow them all to the ends of their natural lives. And one cannot tell if an animal that is not recovered has died or merely moved elsewhere.

Another serious problem for both methods arises from the great instability of all weasel populations. In habitats in which productivity and density vary greatly from year to year, the young entering the population vastly outnumber the breeding adults in good seasons but scarcely match them, or even fail to appear altogether, in poor seasons. Analyses of survival and longevity may then need to separate data from high and low productivity years.

Yet another important source of confusion can be introduced by variation in the season of the year chosen for sampling. Consider two populations, one sampled soon after the end of the breeding season and the other in winter. The first will probably show a far higher proportion of young to adults than the second, because in late summer many more newly independent young are about than if sampling is delayed until overwinter mortality has begun to take its toll. Yet the dynamics and productivity of the two populations could easily be the same.

Finally, all data on age distributions derived from trapping can sample only weasels that survive at least to trapable age; they do not account for young that die at or before birth or in the den. Yet pre-independence mortality is probably the most important factor controlling the age structure of stoats and longtails (Chapter 10). If the total number of weasels born could be known, estimates of the mean ages at death would be much lower, and of the mortality rate of the first year class much higher, than is reported in the current literature. Omission of this information is not serious provided one remembers that all such statistics apply only from the age of independence.