Life tables have also been calculated by the second method, using dead stoats collected in the course of pest control operations in National Parks in New Zealand (Powell & King 1997) and on game estates in Britain (McDonald & Harris 1999). This method also has its problems, but if they can be overcome, a life table can be developed in a shorter time. The validity of the technique depends entirely on three prerequisites:

1. An unbiased method of sampling the population, so that weasels of all ages and both sexes are fairly represented (or if not, that the differences between these groups are at least constant). In addition, sampling effort must either be constant year round or have seasonal variations that are understood and incorporated into the calculations (McDonald & Harris 1999).

2. A reliable method of determining the ages of the dead animals.

3. Confirmation of the important assumption that the distribution of ages among the animals trapped is the same as in the living population (the fact that the animals withdrawn from the population were killed to find out their ages is irrelevant [Caughley 1977:93]).

In a life table calculated this way, the data can be shown as frequency distributions: so many dead animals of age x, so many different ones of age y, so many of age z, and so on. The difference between this type of table and the other is that this one represents the ages of the standing crop (all the members of that population alive at one time), whereas the other represents the life spans of individuals observed over several years.

Because the rate of increase in a real population of weasels is never zero in any one year, and the population age structure varies greatly from year to year, the figures calculated by this method are averages for several cohorts of several years and are only approximate for any given cohort or year. On the other hand, such life tables for weasels collected in different places are remarkably consistent, suggesting that they do represent the general pattern quite well. If the fact that they are only averages is not forgotten, they have their uses.

A preliminary attempt to construct a general life table for stoats, published in the first edition of this book, was calculated by the late Graham Caughley in 1987 by combining samples from three beech forests in New Zealand. In later analyses we (Powell & King 1997) recalculated these data by study site, allowing for the massive annual variations in fertility and age structure (see Figure 10.4; Chapter 10) by compiling two separate life tables representing cohorts of young produced in years of high versus low productivity. A typical result is shown for samples from the beech forest in the Eglinton Valley, Fiordland National Park, New Zealand (Table 11.2 A,B).

A: Eglinton Valley, southern beech forest, seedfall years with abundant |
mice | |||

(159 stoats) | ||||

Proportion |
Mortality |
Survival | ||

Number of |
alive at start |
rate at that |
rate during | |

Age class |
cohorts sampled |
of age class (lx) |
age (qx%) |
age class (px) |

3 months-1 year |
1 |
1.00 |
92 |
0.09 |

1-2 years |
5 |
0.09 |
62 |
0.38 |

2-3 years |
0.03 | |||

B: Eglinton Valley, southern beech forest, nonseedfall years with few mice | ||||

(147 stoats) | ||||

Proportion |
Mortality |
Survival | ||

Number of |
alive at start |
rate at that |
rate during | |

Age class |
cohorts sampled |
of age class (lx) |
age (qx%) |
age class (px) |

3 months-1 year |
3 |
1.00 |
73 |
0.27 |

1-2 years |
4 |
0.23 |
40 |
0.60 |

2-3 years |
2 |
0.14 |
31 |
0.69 |

3-4 years |
2 |
0.10 |
71 |
0.29 |

4+ years |
0.07 | |||

C: Pureora, podocarp-hardwood forest with few mice, |
all years (55 stoats) | |||

Proportion alive |
Mortality |
Survival | ||

alive at start |
rate at that |
rate during | ||

Age class |
Number alive |
of age class (lx) |
age (qx%) |
age class (px) |

3 months-1 years |
31 |
1.00 |
76 |
0.24 |

1-2 years |
13 |
0.24 |
42 |
0.58 |

2-3 years |
3 |
0.14 |
42 |
0.58 |

3-4 years |
3 |
0.08 |
38 |
0.62 |

4-5 years |
3 |
0.05 |
60 |
0.40 |

5-6 years |
2 |
0.02 |

Animals born in each annual cohort were sampled over several years. The distribution of the ages of the dead animals in the total sample is taken to represent the average distribution of ages of the living animals in the population. The sexes are pooled, because their age distributions did not differ in analyses controlled for seedfall. The Eglinton population had been trapped for several years preceeding the dates of sampling; the Pureora population had not. Powell and King (1997) also gave similar tables for two other beech forests, including one (Hollyford) never previously trapped. For cautions, see text. (King et al. 1996; Powell & King 1997.)

Animals born in each annual cohort were sampled over several years. The distribution of the ages of the dead animals in the total sample is taken to represent the average distribution of ages of the living animals in the population. The sexes are pooled, because their age distributions did not differ in analyses controlled for seedfall. The Eglinton population had been trapped for several years preceeding the dates of sampling; the Pureora population had not. Powell and King (1997) also gave similar tables for two other beech forests, including one (Hollyford) never previously trapped. For cautions, see text. (King et al. 1996; Powell & King 1997.)

The results show clearly the enormous mortality of the young stoats, especially in the high-productivity years during a mouse irruption after a seedfall. In the Eglinton Valley, 92% of young stoats born in a seedfall year died before the age of 12 months, and the whole cohort was gone within 3 years. This was not an exceptional observation; in two other beech forests the mortality rate of first-year stoats after mouse peaks was 91% and 92% (Powell & King 1997).

By contrast, in the intervening years when fewer stoats were born, first-year mortality was lower (but still averaged 73% over four cohorts)—and the remaining few older adults enjoyed relatively good survival over the following 2 years (mortality 30% to 40%). From the fourth year onward, their mortality rate started climbing steeply again. In the only nonbeech forest sampled, at Pureora in the central North Island, the density of mice and the density and population dynamics of stoats were much like Fiordland in nonseedfall years (Table 11.2).

The New Zealand data document a drastic reduction in the prospects of survival of the young stoats born during the stoat "plagues" that coincide with a mouse irruption in these forests (Figure 11.4). If that is true, it leads to the prediction that the oldest members of the population should comprise mostly individuals born in low-density years. When we tested that prediction from our data, we found that, of 70 stoats over 3 years old, only seven males and 13 females had been born in the highly productive mouse peak years (Powell & King 1997).

Figure 11.4 The very large numbers of young stoats born after a heavy seedfall in New Zealand beech forests have a consistently lower survival probability than those born in the years between seedfalls. Stoats born in postseedfall years have solid circles and lines; stoats born in nonseedfall years have open circles and dashed lines. (Redrawn from Powell & King 1997.)

Figure 11.4 The very large numbers of young stoats born after a heavy seedfall in New Zealand beech forests have a consistently lower survival probability than those born in the years between seedfalls. Stoats born in postseedfall years have solid circles and lines; stoats born in nonseedfall years have open circles and dashed lines. (Redrawn from Powell & King 1997.)

McDonald and Harris (2002) made the first large (n = 822) collection of stoat carcasses from British gamekeepers. On the farmlands and moorlands they sampled, there are of course annual variations in food supply for stoats, but they seldom reach the feast-or-famine proportions typical of New Zealand beech forests. On the other hand, trapping by gamekeepers is concentrated in spring and early summer, when removal of stoats can best benefit nesting game birds (Chapter 12). Consequently, the samples were affected less by annual variation in resources than by the rather different problem of seasonal variation in trapping effort.

The data show the patterns of age-specific and annual mortality typical of stoats in general, and also, not surprisingly, that the mortality rates of both sexes and all ages of these particular stoats were much higher in spring than at other seasons (Table 11.3). The very youngest stoats just out of the den (0 to 3 months old) comprised the largest single group of stoats caught, but their mortality rate was proportionately much lower (10% to 14%) than that of the smaller group of 12- to 15-month-olds caught in the same season (83%). For both age groups, the normal spring danger period (when rodents are at their seasonal low and before young rabbits become available) was artificially made worse by the management strategy of the gamekeepers.

McDonald and Harris also used contemporary analytical tools to explore parameters not previously considered, such as which of several possible factors

Male stoats Female stoats |
Male weasels |
Female weasels | |

1968-1972 |
(King 1980c) | ||

Summer |
— — |
14 |
15 |

Autumn |
— — |
16 |
11 |

Winter |
—— |
21 |
22 |

Spring |
— — |
65 |
58 |

First year |
80 |
75 | |

1996-1998 |
(McDonald & Harris 2002) | ||

Spring |
10 14 | ||

Summer |
13 27 |
15 |
8 |

Autumn |
13 30 |
24 |
30 |

Winter |
39 40 |
16 |
16 |

Spring |
83 83 |
73 |
100 |

First year |
59 74 |
85 |
97 |

1. Ages estimated from carcasses. Median birth dates assumed to be April 1 for stoats and June 1 for weasels. Three monthly seasons for the two species starting from the median birth dates were therefore defined differently: For stoats the first month of each season was July (summer), October, January, and April; for weasels, June, September, December, and March. (McDonald & Harris 2002; King et al. 2003a).

1. Ages estimated from carcasses. Median birth dates assumed to be April 1 for stoats and June 1 for weasels. Three monthly seasons for the two species starting from the median birth dates were therefore defined differently: For stoats the first month of each season was July (summer), October, January, and April; for weasels, June, September, December, and March. (McDonald & Harris 2002; King et al. 2003a).

most strongly influence population growth rate. Stoats and weasels in general have very unstable populations, early litters, and short lives (King 1983a; McDonald 2000). For them the population rate of increase varies every year, from a positive value when the population is rising, to a negative one when it is declining.

Species in which the rate of increase (r) fluctuates between high and low values tend to "live fast and die young," as McDonald put it (2000). They are capable of very rapid but temporary variations in density, and are very resistant to control. Stoats and common weasels have long been regarded as so-called r-strategist species (Chapter 14) but until McDonald and Harris' analysis, r had never been calculated for either. McDonald and Harris constructed a mathematical model that suggested that r averaged -0.05 for the British stoat population, at least during the years they sampled.

This negative figure does not mean that British stoats are constantly declining. All such population models demand some conventional but unrealistic assumptions, such as that survival is independent of density and that the population analyzed was closed. Nevertheless, even though we know those assumptions are not accurate in fine detail, the model is useful because it has shown for the first time that the single factor most critically influencing r for stoats was the survival rate of second-year females from April to June, the period when they produce and suckle their first litters.

If a mother stoat dies before her young can kill for themselves (at about 10 to 12 weeks old; see Table 9.3), it is probably safe to assume that the whole litter will be lost as well, so trapping females at that time of year has a disproportionately large effect on the population. If the model's survival rate for that group of females was adjusted by only a small amount, from 48% to 54%, r was lifted from slightly negative to an even zero, and presumably in years or in places where the survival of this group is high, r becomes positive. When those same females are also weaning large litters, because a period of great food abundance has permitted substantial reduction in juvenile mortality, a positive trend could easily turn into an irruption. Take, for example, the irruption that followed the introduction of stoats to Terschelling, off the coast of the Netherlands. The island teemed with water voles when, in 1931, about nine stoats were released on the island. According to van Wijngaarden and MSrzer Bruijns (1961), the colonizing group had multiplied to at least 180 by 1934.

Chapter 10 provides plenty of confirmation that the r statistic for stoat populations frequently does oscillate from positive to negative, producing the unstable numbers typical of stoats everywhere. Clearly there are times when the dynamic equations of local stoat populations can absorb high rates of mortality, both natural and imposed, and times when they cannot. In patchy environments there will always be a mosaic of breeding opportunities for stoats. The high dispersal capabilities of their offspring (Chapter 9) ensure that there is always a supply of young to migrate across the landscape eager to recolonize vacant patches after every local extinction (McDonald & Harris 2002).

Another model explored the dynamics and productivity of stoat populations in New Zealand beech forests (H. Wittmer, R. A. Powell, and C. M. King, unpublished), using the survival data calculated by Powell and King (1997). Wittmer et al. developed age-specific, population projection matrices for stoats living through each phase of a typical beech masting cycle. During a seedfall year, the growth rate of the model population was strongly positive, because the high numbers of mice in late winter enabled the females carrying a normal-sized litter (conceived before the seedfall) to attain such good body condition that intrauterine and nestling mortality was minimized. Therefore, very large cohorts of young were produced at the beginning of the postseedfall year (see Figure 10.4). But these young females emerged (already fertilized) just as the mice were disappearing, and few of them survived to their first littering season as 1-year-olds the next year; those that did lost most or all of their young. Hence, population growth during the postseedfall year was strongly negative. During the transition years between each crash and the next seedfall year, the population was nearly stable. Wittmer et al. linked successive population matrices together to calculate elasticities of the different vital rates at different points in the cycle. This analysis found that the fertility and the survival of 0-1 year old stoats in the post-seedfall year both had great influence on the dynamics of the population. In addition, the analyses demonstrated that the rate of population growth through the cycle differed from that predicted from the stages of the cycle analyzed in isolation. These results illustrate the critical importance of understanding how stages of a population cycle are linked together.

Was this article helpful?

## Post a comment