The seasonal distribution of births

The overall median date of birth for the Orongorongo possums (18 May) was relatively late compared with many populations studied elsewhere. Most medians for autumn births recorded in Australia fell in April or early May (Dunnet 1956, 1964; Smith et al. 1969; How 1972; Winter 1976), while in New Zealand medians in April or early May were recorded for Banks Peninsula (Gilmore 1966; Jolly 1976), Tokoroa (Clout 1977), Ashley (Warburton 1977) and the Taramakau (Bamford 1972) ; on Kapiti births were earlier in 1975 (Bell and Atkinson 1976) but generally not so in later years (R.E. Brockie, pers. comm.).

As noted by Kean (1971), Crawley (1973) and others, the autumn birth period of the possum is often supplemented by another in spring that involves fewer females. In Australia spring births occurred in most populations studied and reached 30% of births in Canberra (Bolliger 1942; Dunnet 1956, 1964; Lyne and Verhagen 1957; Smith et al. 1969; How 1972; Winter 1976). Spring births have also been widely reported in New Zealand and, as in Australia, appeared most prevalent in populations where autumn births were earlier e.g. Banks Peninsula (Tyndale-Biscoe 1955; Gilmore 1966, 1969; Jolly 1976), some of the Hokitika (Boersma 1974) and at Tokoroa (Clout 1977). Spring births were virtually absent in the Orongorongo Valley over 1966-75 (Fig. 2) only three occurring, two in 1971 when, significantly, autumn births were unusually early (median 19 April).

The same populations with early autumn breeding and a high proportion of spring births were also those in which some females evidently reared young in both autumn and spring. Kean (1971) concluded that such animals page 118 occur only when densities are low and food supplies are plentiful. However, he considered such conditions would not always result in double-breeding, but potential expansion of population was a common factor. Boersma (1974) indicated some possums in the Hokitika catchment with spring births had bred twice; spring births there only occurred in populations with high fat reserves and high asymptotic weights.

In the Orongorongo Valley, Kean (1971) expected sustained gin-trapping during 1953-61 would lead to the emergence of double-breeders in better locations, but this did not evidently occur, even when possums near the field station were given supplementary food - though an isolated female showed oestrous behaviour in November (Kean unpubl.). Kean (1971) concluded a high incidence of double-breeders required selection for appropriate genotypes, and could not be induced somatically. Nevertheless, the present study illustrates the possible importance of localised environmental factors in determining the reproductive capabilities and success of the Orongorongo female possums; such environmental constraints may well have still been operating to some degree when the population had formerly been reduced in Kean's experiment.

Age of first breeding

Most Orongorongo females did not have young until at least 2 years old; the few records of yearlings breeding occurred only in 1969 (2) and 1971 (5). This contrasts with many other study areas where yearling females usually have young, as at Canberra (Dunnet 1956, 1964), in New South Wales (Smith et al. 1969, How 1972), on Banks Peninsula (Gilmore 1966; Jolly 1976), in parts of the Hokitika catchment (Boersma 1974) and in Ashley Forest (Warburton 1977).

Survival rate of pouch-young

Despite their close association with the female, only about half of the pouch-young (52%) survived in the Orongorongo Valley over 1966-75. This ranks as a poor survival rate compared with some other reports. For instance in Canberra Dunnet (1964) found survival in the pouch was very high, while Kean (1971) concluded in an earlier study that most Orongorongo females had reared young. There was marked variation in pouch-young survival between years, areas and age-classes. In area A no yearling females reared pouch-young, while in area B neither yearling nor two-year-old females did so (Tables 1,2,4). Adult females raised page 119 only 30–50% of their pouch-young in area B, while in area A there was marked annual variation in pouch-young survival ranging from 30% (1975) to 90% (1971).

Recruitment rate of young

The inter-correlation of annual measures of breeding performance in the Orongorongo Valley over 1966-75 is emphasised in this paper; similar relationships are evident when the breeding productivity is compared between different study areas. Broadly, areas with earlier autumn breeding, such as Canberra and Banks Peninsula (Tyndale-Biscoe 1955; Dunnet 1956, 1964; Gilmore 1966, 1969; Jolly 1976) are those with highest rates of spring births and double-breeding, with more yearling breeders and with high survival of pouch-young, so the resultant breeding productivity is much higher. Kean (1971) indicated such populations occur where food supplies are relatively good and densities are low; by contrast Orongorongo possums are at higher density (Crawley 1973), are depleting their food species (Mason 1958; Fitzgerald 1976, 1978; Meads 1976) and are subject to marked annual variation in productivity, which in general is low.

Assuming 20% of females are between one and two years old, the estimated annual production of independent young was 1.16 per female in Canberra (Dunnet 1956, 1964; Brockie et al. 1979). The overall production rate for Orongorongo females over 1966-75 (derived from data in Table 1) was only 0.37 young per female. This has interesting demographic implications when related to available data on mortality of the study population.

Brockie et al. (this symposium) estimate survivorship rates from a sample of 103 possums found dead in the study area over 1966-74. These data are combined with breeding data from Table 1 in Table 9; age classes follow Brockie et al. except all animals 3 years and over are combined as 'adults'. Using a Leslie projection matrix (Leslie 1945, Lefkovitch 1965, Williamson 1972) on the available data, the finite rate of natural increase (λ) is estimated as 1.01 (using either assumed sex-ratio and with either the unadjusted mortality series or with mortality series smoothed by fitting quadratic curves - see Snedecor & Cochran 1967). This estimate indicates the population is producing just enough young to counter mortality losses in older animals. However survival page 120

Table 9. Summary of annual recruitment and survival of possums, Orongorongo Valley 1966-75.

AGE CLASS1 (years)	BREEDING RATE2	MALE SURVIVAL1	FEMALE SURVIVAL1	EXPECTED NO. MALE OFFSPRING3	EXPECTED NO. FEMALE OFFSPRING3(FERTILITY RATE)
0–1	0	0.893 (0.821)	0.851 (0.831)	0 (0)	0 (0)
1–2	0	0.900 (0.876)	0.875 (0.884)	0 (0)	0 (0)
2–3	0.303	0.911 (0.914)	1.000 (0.922)	0.155 (0.151)	0.148 (0.151)
Adults	0.449	0.805 (0.831)	0.833 (0.849)	0.230 (0.225)	0.219 (0.225)

page 121 rates may be overestimated; losses between the end of pouch-life (170 days) and the young becoming free-ranging (and at risk of 'entering' the life-tables) are not included in estimates of recruitment rates. Thus it appears that the population may barely be capable of maintaining its numbers without net immigration from outside, although in reaching such a conclusion some inadequacies of the data need to be borne in mind (see Brockie et al. this symposium, also earlier Discussion).

Adult weight trends

Data on seasonal weight trends in adult females (Fig. 6) suggest more successful breeders increase their fat reserves during autumn and into winter, though a decline occurs in late winter or spring, probably related to the maximal demand of lactation (Gilmore 1966, Kean 1971, Crawley 1973). Kean (1971, unpubl.) in 1953-61 and Crawley (1973) in 1966-68 found fat reserves/mean weight increased in winter. Kean (unpubl.) described 72% of August females in 1953 as well conditioned (as opposed to lean) but by September the proportion dropped to 47% (when of 34 animals 5 had no fat, 13 a trace of fat, 6 were in good condition and 10 were very fat). It is evident seasonal weight trends varied between years; so although on average winter weights increased during 1966-68 (Crawley 1973) over ten years (1966-75), different years tended to cancel each other out and overall there was no significant seasonal effect.

Gilmore (1966) found on Banks Peninsula (where spring births occurred) female weights and fat reserves peaked in autumn and remained relatively high in June and July, and a similar pattern was evident there more recently (J.N. Jolly, pers. comm.). Bamford (1970) found females from page 122 the lower Styx Valley, Westland, had highest fat reserves in June compared with March, September and December, while in exotic pine forest near Tokoroa female fat reserves and weights again peaked in winter (M.C. Clout, pers. comm.). Thus, despite some minor regional variation, it appears females, where possible, lay down fat reserves in autumn and early winter, but that these reserves are utilised during late winter or early spring when pouch-young approach their periods of maximum growth (Kean 1975), and demands of lactation are higher (Gilmore 1966, Kean 1971, Crawley 1973). The present study suggests that some females are unable to metabolise fat reserves in this way and either do not have young (perhaps because they do not return to oestrous in autumn) or are unable to sustain a developing and suckling pouch-young through the winter.

Pouch-young growth

Orongorongo pouch-young attain relatively low annual mean weights (229–439 g) by the end of pouch-life. Possums near Canberra (Dunnet 1956) were much heavier (ca. 900 g) at 170 days; Lyne and Verhagens' (1957) nomogram indicates a weight of almost 600 g for other Australian data; Gilmore (1966) recorded a mean of 860 g for Banks Peninsula possums at 5–6 months. Kean (1975) found captive possums in the Orongorongo area over 1953-61 weighed 240–580 g at the end of pouch-life (ca. 170 days). His calculated growth rate of wild pouch-young, based on a field collection of 811 individuals, evidently underestimated their mean weight at 170 days; the estimate (198 g) was lower than any comparable figure for 1966-75 and is hard to reconcile with a juvenile weight of 850–900 g in November-December 1953 (Kean 1975 - Fig. 5). These juvenile means were higher than those in this study which averaged (± S.E.) 464 ± 17 g in November and 648 ± 31 g in December.

Juvenile growth

Growth of the juvenile over the months after leaving the pouch also varied markedly between years as evident from mean weights of yearlings which ranged from 0.83–1.55 kg in area A and 1.17–1.45 kg in area B (Table 8). While no yearling females bred in area B, some did in area A in the two years when mean weights were highest (1969, 1971). Higher growth rates and yearling weights were noted in other studies, for instance by Lyne and Verhagen (1957) and Dunnet (1964) in Australia and by Gilmore (1966) in New Zealand.

Population density

Annual estimates of population density in area A over 1966-75 were compared with breeding and weight data to see if any density-dependent trends were evident. The population density was approximately estimated by determining the number of possums known to be alive in the trapped area each year; this ranged from 150 (1975) to 217 (1967). There were no significant correlations between the current (annual) population density and the annual weight and breeding performance (Fig. 10). Had increasing density depressed weights and breeding output, then negative correlations with adult birth rate, pouch-young survival rate and adult winter weight would be expected, while the date of birth would be positively correlated. There is evidence of such trends in some of the graphs if one ignores the less typical years of 1971 and 1975.

If the population density the following year is compared with certain breeding and weight data then significant correlations did occur (Fig. 11). Contributing to this delayed effect is the fact that most young were not trapped as free-ranging animals until the calendar year following birth. It is to be expected that a year of improved breeding success would contribute more young to the population. This, combined with better adult winter survival in such years, seems to be the basis for the observed correlations.

Fig. 10. The current annual population level (minimum no. known alive) in relation to annual breeding and weight parameters, area A (1966-75). Arcsin transformations for percent adults with young and percent rearing pouch-young. Correlation coefficients (r) and significance levels (p) are given for all 10 years and (in brackets) for the 8 years excluding the 'extreme' years of 1971 and 1975.

Fig. 11. Linear regressions and correlations between certain annual breeding and weight parameters and the number of possums known to be in the population the subsequent year, area A (1966-75). Arcsin transformation for percent adults with young.

The area B study was initiated with the aim of investigating the effects of an experimental reduction in population density on the surviving population. However, the contrasts between the two populations proved of sufficient interest to postpone the experiment; but it would still be interesting to monitor the animals' response to reduced density and observe its effect on productivity. Kean (1971) has already discussed some aspects of reducing density in the Orongorongo population.

Food supply

Annual variation in the quantity and quality of possum foods could be important in determining their condition and breeding success. One component of their diet known to fluctuate from year to year comprises the fruits, berries and seeds of various forest plants (Fitzgerald 1976, 1978; Waddington et al. 1980; M.J. Daniel, pers. comm.). Such food supplements the predominantly leaf diet of the possum population. While data on the annual production of such food items were not available for earlier years of study, it is interesting that the year 1971, notable for the high breeding success in area A, was also a year of abundant fruit and seed production there. Despite limited data, correlations were evident between some indices of fruit and seed production and some weight and breeding parameters (Fig. 12).

It was fortunate that data on hinau fruit were available for analysis; while certainly an important food of possums, it is not the only fruit taken by them (Mason 1958, Fitzgerald 1976, 1978). Pigeonwood, for instance, was eaten by many possums early in 1971 when it fruited abundantly (Fitzgerald 1976; pers. obs.).

Among the range of plants eaten by possums there is marked variation in reproductive cycles and probably in the factors influencing fruit production. Also such fruits are not the only variable components in the diet - possums take leaves and flowers of a variety of species, as well as invertebrate foods at times (Fitzgerald 1976, 1978; pers. obs.). This variation in the quality and quantity of food available for the possums therefore seems to have been important in influencing annual fluctuations in weight and breeding performance, especially in years like 1971 when area A possums bred most successfully.

Comparison between study areas

There were differences in the composition and condition of vegetative cover in the two study areas; for example virtually all northern rata trees, a known food species, were dead in area B by 1971, unlike area A (Meads, 1976; pers. comm.). More detailed comparative vegetation surveys were not completed until after 1974, when live-trapping in area B was halted (A.E. Fitzgerald, pers. comm.). However, it was apparent that a number of food species apart from northern rata (e.g. kamahi) were absent or reduced in numbers in area B, possums having heavily browsed them (see earlier description of study areas). Supplementary foods such as buds, flowers, fruit and seeds occurred commonly in the diet of possums in area A (Fitzgerald 1976); a comparison of fruit, flowers and seeds in the litterfall suggests area A produces a much greater annual crop of such foods compared with area B (Waddington et al. 1980). Not only is area A vegetation more diverse but there are marked differences in those species whose fruit is taken by possums (A.E. Fitzgerald, pers. comm.). The 4 species probably contributing most of the fruit eaten in area A - hinau, pigeonwood, pate and supplejack -bear ripe fruit over autumn and winter. Only pigeonwood and supplejack are common in area B, most of the other species of fruits eaten there being available earlier in the year (up to April) e.g. kaikomaka, kawakawa, karaka Corynocarpus laevigatus and mahoe (Waddington et al. 1980).

Thus differences in the seasonal weight trends and breeding productivity between areas A and B could well reflect differential food supplies, differing not only in the leaf component of the diet but in the timing and availability of supplementary foods.

Weather

Since continuous weather records were not available from the study areas over 1966-75 N.Z. Meteorological Service data from Kelburn, Wellington and the Orongorongo Weir are used to summarise annual and seasonal fluctuations in rainfall and temperature over the study period (Tables 10 and 11). Kelburn recorded about half the amount of annual rainfall as the study areas, while at the Orongorongo Weir it was about 400 mm (14%) more. Over 1969-71 mean winter and summer temperatures at Kelburn were milder than area A by 2.6°C and 1.2°C respectively.

Fig. 12. Linear regressions and correlations between the annual crop of hinau fruit and various annual breeding and weight parameters in area A.

While hinau is but one of the fruits eaten by possums, there is a correlation between the annual hinau crop (no. fruits/m²) and the annual crop of all fruits and seeds collected in the area (measured as kg/ha, dry weight - see Daniel 1975 for methods). Although data are limited, this suggests hinau cropped heavily in concert with some other species in years like 1971 (see Figure, lower right based on 1971-75 data).

For percent adults with young arcsin transformation.

Annual mean daily temperatures at Kelburn (Table 10) were highest in 1971 (13.5°C) and 1969 (13.2°C) and lowest in 1966 (12.2°C), 1967 and 1972 (12.4°C).

The summer and autumns of 1975 and 1971 were especially mild, while those of 1969, 1967, and 1966 tended to be coolest (Table 11). The winters of 1971 and 1970 were relatively mild, while those of 1972 and 1969 were cooler. Spring temperatures were highest in 1972 and lowest in 1968, 1966 and 1975 (Table 11).

Rainfall data show the years 1974, 1966 and 1968 to be the wettest, while 1969 was especially dry (Table 10). Seasonally the summers of 1967 and 1966 were wettest; the 1973 summer had the lowest rainfall, while the summers of 1971 and 1974 had the longest dry periods (fewer raindays). Heaviest autumn rainfalls occurred in 1974 and 1966, the most sustained rain occurring in 1968. Conversely, lightest falls occurred in the autumns of 1971 and 1973, with least sustained rain in 1967, 1969 and 1972. In winter most rainfall occurred in 1975 but was most sustained

^* see Table 1 for calculation of percentage deviations

page 131 in 1972. The driest winter was 1969. Spring was wettest in 1968, 1971 and 1974 and driest in 1969 and 1966 (Table 11).

An analysis of Kelburn weather records and possum weights showed adult female summer and autumn weights were correlated with prevailing weather; the animals weighed more in drier, warmer seasons. Also the mean daily summer temperature was correlated with mean weights one year later - possibly a spurious result, although an indirect, delayed effect through, say, the growth of possum food plants could be indicated. Lack of some local meteorological data made study of immediate weather and possum weight and breeding data difficult; very poor, wet weather reduced the numbers of possums caught, and a sustained period of bad weather could curtail feeding and reproductive activity and so contribute to reduced breeding; such may have been the case in the 1968 autumn and winter (refer Fig. 7); further, many of the possums found dead or comatose were trapped in cold and wet weather. Thus despite limitations in data, it appears that for 1966-75 weather fluctuations probably directly or indirectly influenced the possums and their food supply, and so may ultimately have influenced possum weight and breeding success.

Pathological factors

Autopsies were limited to those animals found dead or dying and little information was obtained on diseases and parasites.

At a time of heavy adult mortality in winter 1972 the cestode tapeworm Bertiella trichosuri Khalil was found in 5 dead possums examined; a further 2 cases occurred in winter 1973 (A.J. White, pers. comm.). Clark (1977) reports lower weights and low birth rates in females infected with this parasite in Taranaki. Some Orongorongo possums had heavily abraded fur due to irritation from the ectoparasite mites, Atellana papilio and Trichosuro laelaps crassipes; animals in poor condition were most heavily infected, and were severely scratched on the rump. Whether poor-conditioned animals were more prone to infection or whether the parasites contributed largely to their poor condition is not known.

Bovine tuberculosis Mycobacterium bovis, reported from New Zealand possums (Ekdahl et al. 1970; see also Julian, this symposium), was not observed in the study area during a 1976 survey (A.J. White, pers. comm.); advanced stages of infection characterised by external lesions at the lymph nodes would be evident in regularly trapped animals.

High infection rates of leptospirosis have been found in New Zealand possums (de Lisle et al. 1975; Hathaway, this symposium). A 1975 survey of urine samples from the Orongorongo study population revealed an infection rate of 10 percent (R.E. Brockie, pers. comm.), but any relationship between leptospore infection and breeding performance has yet to be established.