



Chapter IITI.--Characteristics of Avian Life Tables 
Quality of primary data 
The primary data in any life table may be regarded as the 
original information collected in the field. Even in the work on 
man, the quality of these may not be all that is desired. Census 
tabulations have been found, for instance, to show age-groupings favor- 
ing mltiples of 5 or 10. They are also invariably deficient in giv- 
ing the number alive in the first year of life (Hill 1936). These 
weaknesses are minor compared to the ones encountered in avian life-~ 
table work. 
The striking defect in the quality of primary data available 
on bird populations lies in the absence of facts either on the numbers 
alive or on the numbers dying. The two sources of data have yet to be 
jointly available for work on the same population, although they are 
now separately available for a number of species. The defects in the 
quality of data arising from a retrapping program have been described 
in Chapter II. The quality of the mortality data may be examined here. 
Two questions regarding it are pertinents 
Is a banded sample of a bird lation representative of the 
species? In some cases, the answer is yes; in others, no. Some species, 
Tike the small songbirds, are banded by hundreds or thousands throughout 
their breeding range. When nestlings are banded in this way, the life- 
table results may be fairly safely regarded as originating from a ran- 
domized sample of the species! population. In other species, the banded 
birds may all come from a single region or even a single colony. Life- 
table results in such circumstances should not be construed as holding 
for a continental population, although they may nonetheless be very 
valuable. Special problems may or may not be encountered when life- 
table results are based on trapped samples. Game species that in any 
way become conditioned to baited traps should be suspect in this con- 
nection, at least until better data are available. A great many water— 
fowl have, for instance, been banded by the U. S. Fish and Wildlife 
Service on its network of refuges throughout the country. At the 
present time, no one knows if these birds have better or poorer sur- 
vival rates than nonrefuge birds banded at random as fledglings on 
scattered breeding grounds in Alaska and Canada. The survival-rate 
implications of a goose refuge, described by Elder (196), might be 
mentioned in this connection. Another variable seems to present itself 
when we consider life-table results based on a short period of banding, 
in contrast to those arising from a banding program spread over many 
years. The former may give survival rates peculiar to an increasing 
or decreasing population; in the latter, the rates would tend to be more 
typical of a stable population (although this is not necessarily a safe 
assumption). Some of these variables will be encountered in the life 
tables constructed in Part II. Their overall effects may be small, but 
their presence should caution us to accept the results with reserve. 
27 
