Sampling distortions

The samples considered in this paper were collected in many ways and for diverse purposes. They present a heterogeneous base on which to build any sort of speculative analysis. The main inadequacy is that we know little or nothing of the stability or instability of the populations.

Collections of possums made over several months or years, although useful for investigating, for example, age-specific fecundity or the life expectancy of older animals, are of little use in reconstructing the recent history of the population. So many events and processes, acting separately or in concert, overtake the population during the period of collection as to make interpretation difficult or impossible.

'Instantaneous' samples make it possible to date the effect of recent events on the age structure with some accuracy.

The patterns of age structure were more regular in the female than in the male component in most samples. Males showed greater variations from the expected patterns, presumably because they were less sedentary than the females and because of their greater vagility - young males probably moved quickly to fill depopulated areas and distorted samples because they were more vulnerable to traps, poisons and guns.

Seasonal change in age structure

Two main patterns of age structure have emerged - that prevailing from October to early February, with a predominance of 0–1 year-olds, and a late February to September pattern, with a modal age of 1–2 years.

The 'February' transition comes unexpectedly early as the main birth peak for most New Zealand possums occurs in May (Brockie et al. 1979) and animals should not graduate into the 1–2 year class until May of the following year. The Wainuiomata sample (Fig. 2a) shows that the bulk of the animals had developed their first layer of dental cementum by late February or early March when they were probably 10–11 months old; more known-age yearlings must be page 79 examined to resolve this point and more field samples collected between February and May are required to date the transition in other localities.

Three notable exceptions to these age structure patterns require explanations. The Waitotara (1974) sample has a bimodal pattern peaking in the 0–1 year, and the 2–3 year age classes; and males in the Tokoroa sample of September 1974 have a peak in the 3–4 year age class. Clout (1977) explains the large number of 3–4 year old males in the September Tokoroa sample as being due to an influx of young males to fill gaps left by the clearing and burning operations 4 years earlier. The Waitotara animals were subjected to a drastic control program 3½ years previously and the large number of 2–3 year olds in the later sample is also probably due to the rapid influx of one-year olds to fill the gap. The Copland Valley males reveal the usual October-February age structure but the females are quite anomalous (Fig. 1d) with a predominance of 4–5 year olds - cf. 5 males and 20 females in this age class. Fraser (1979) does not attempt to explain this anomaly but it appears as though the population was disturbed in 1973 or 1974 and that it affected males more than females. Other possibilities are that few female young were produced in 1974-77, or that young females were subject to greater mortality than males during those years.

The large number (58.7% of males and 47.8% of females) of 1–2 year-olds from the Tennyson Inlet sample is not approached by any other sample. The trapped area consisted of a narrow strip of seaside forest at the back of which lay an extensive bushed hinterland. The high proportion of 1–2 year-olds probably resulted from the continued but light trapping over a small area, providing empty living space which was continually refilled with youngsters dispersing in from the hinterland. The ranger on Kapiti Island (Peter Daniel, pers. comm. 1978) reported a somewhat similar pattern. Over 18 months he shot some 130 possums near his house at Rangatira Flat. After the first 100 animals, almost every new animal shot was a young male.

Sex-ratio and age classes

The excess of males in the 0–2 year age class of most samples is probably caused by two factors:

1.	Excess of males in the pouch. Caughley & Kean (1964) confirmed that males slightly outnumbered females in 908 possum pouches, in the ratio of 100 females: 114 males. page 80
2.	Sampling bias. Young animals, especially young males, experience a dispersal phase in the first year or two of life (Dunnet 1964). Males also exploit a larger home-range than females (Ward 1978) so, moving more widely, are more likely than females to encounter traps, poison or spotlight shooters, a point also made by Fraser (1979).

Table 3 shows that females dominated all 12 age classes over the age of two years. Without further pooling the differences in the sex-ratio are too small to sway tests of statistical significance but their consistency is impressive.

Over the age of 9 years females dominate the pooled samples by 38:14 and by 19:3 over the age of 10 years. This excess of females in very old age classes is not exceptional in mammals. Dall sheep Ovis dalli, Orkney voles Microtus orcadensis and humans show the same tendency (Caughley 1966).

The high mortality rate of 2 year-old males may be attributed to the extra hazards they meet while dispersing and their continuing decline after the age of 3 years may be caused by their having to maintain a larger home-range than females with its attendant extra demands. Another possibility is that male possums share the kind of reproductive stress shown by the brown antechinus, Antechinus stuartii, which goes into a general decline after mating. The metabolic rate of the male Antechinus shifts up during the breeding season and the animals move into a negative nitrogen balance in April, unlike the females which remain in a positive nitrogen balance during the winter (Woollard 1971). Our records from the Orongorongo Valley and Kapiti Island show that most males lose considerable weight in the winter whereas females generally maintain or increase their body weight (see Bell, this symposium).

The greater longevity of females does not show up in the sample of animals dying natural deaths in the Orongorongo Valley, despite the four thirteen-year-old females. The small size of the over-eight-year-old samples precludes statistical tests of significance.

Mortality rate and life expectancy

The calculations on mortality rate and life expectancy of possums in the Orongorongo Valley (Table 5) give a spurious appearance of accuracy and may be wide of the mark because (1) the sample size is so small, (2) both sexes are run together in the calculations but there is every possibility that the sexes suffer differential mortality rates, and (3) the table assumes a constant page 81 population density before and during the period of collection. We have evidence, however, that possum numbers rose from about 7 to about 15 animals per hectare in 1972 and fell away again after that date (unpublished data). The calculations therefore give a clue to mortality rates and life expectancy but the details are open to question.

However, Boersma (1974), who also supposed his mixed samples were from a stationary population, obtained rather similar results. He calculated that females suffered high mortality in their first year, a reduced rate between their first and fourth years and an increased rate after the age of four. These similarities between very different populations perhaps give some plausibility to the calculations.

We must concur with Caughley (1974) that the age structure of possum populations is of little value in estimating whether a population is on the increase or decrease. Age-specific birth and death rates, the incidence of double breeding and spring births, the migration or dispersal of certain age classes, overwhelming degradation or improvements in the habitat, control operations, food crop successes or failures, and perhaps predation and disease, shape the pattern of age structure. Add to these factors the bias of most sampling methods and the usual demographic changes which overtake an animal population from month to month, and the hazards of interpretation become apparent.

Age estimation is nevertheless a valuable exercise as it throws light on the age of maturity, age-specific reproductive performance and mortality, the potential and actual rates of increase, the age-specific effects of diseases and control operations.