280
H . K L O M P
Two other years are worth considering, namely 1962 and 1963.
Generation mortality is then also higher than expected according to
larval mortality, and in both years this results from a high pupal mortality, but the causes differ. In the 1962163 season the fraction of
pupae killed by parasites is in the normal range, but the post-census
mortality of pupae is high as a result of an abnormally high density of
the larvae of the predatory beetle, Athous subfuscus. In the 1963/64
generation, on the other hand, the parasites Blondelia and Poecilostictus
cause a high pupal death rate (Tables XV, XXIII, and Fig. 31).
Figure 31 further demoiistrates that generation mortality is not
affected by the minor fluctuations of the sex ratio, and to an insignificant extent only by the density induced variability of the reduction of
fecundity. The sex ratio is a factor on an intermediate level, but ha,rdly
fluctuates, and can thus be expected to have no effect. Fecundity reduction varies more, but is ineffective as a result of the low level on which
the fluctuations occur (p. 277).
To sum up, it is shown that the fluctuations of the population density
are primarily controlled by larval mortality changes, with occasional
secondary effects of egg and pupal mortality superimposed upon them.
2. Age-interval Mortalities and their Contribution to Density
It can be objected that the conclusions reached in the foregoing section are supported only by a visual comparison of the graphs of Fig. 31.
To meet this objection we applied the method of key-factor analysis
described by Morris (1959, 1963b) to the same population data. This
method starts from the fact that the density of a definite stage, e.g.
egg density, is determined in succession by the egg density of the previous generation, the mortality of the eggs, the mortality of the larvae,
the mortality of the pupae, and so on. To be informed about the effect
of the separate components, the correlation between the egg densities
of two successive generations is first studied, moving from the first to
the last generation for which data are available. Secondly, to learn the
effect of egg mortality, the correlation between the density of first instar
larvae and the egg density of the next generation is studied, and so on.
The quantitative formulatiojl of this method starts from the function
Et+i = Et . sl. S, . . . . . . s I 2 . F
where Et = egg density of generation t
= fraction surviving in the successive age intervals (s, = 1-p,, where
= fecundity (the maximum fecundity in this study, being 216 egg;gs/
8,
F
pi is the fraction apparent mortality in the relevant interval)
female )
H . K L O M P
Two other years are worth considering, namely 1962 and 1963.
Generation mortality is then also higher than expected according to
larval mortality, and in both years this results from a high pupal mortality, but the causes differ. In the 1962163 season the fraction of
pupae killed by parasites is in the normal range, but the post-census
mortality of pupae is high as a result of an abnormally high density of
the larvae of the predatory beetle, Athous subfuscus. In the 1963/64
generation, on the other hand, the parasites Blondelia and Poecilostictus
cause a high pupal death rate (Tables XV, XXIII, and Fig. 31).
Figure 31 further demoiistrates that generation mortality is not
affected by the minor fluctuations of the sex ratio, and to an insignificant extent only by the density induced variability of the reduction of
fecundity. The sex ratio is a factor on an intermediate level, but ha,rdly
fluctuates, and can thus be expected to have no effect. Fecundity reduction varies more, but is ineffective as a result of the low level on which
the fluctuations occur (p. 277).
To sum up, it is shown that the fluctuations of the population density
are primarily controlled by larval mortality changes, with occasional
secondary effects of egg and pupal mortality superimposed upon them.
2. Age-interval Mortalities and their Contribution to Density
It can be objected that the conclusions reached in the foregoing section are supported only by a visual comparison of the graphs of Fig. 31.
To meet this objection we applied the method of key-factor analysis
described by Morris (1959, 1963b) to the same population data. This
method starts from the fact that the density of a definite stage, e.g.
egg density, is determined in succession by the egg density of the previous generation, the mortality of the eggs, the mortality of the larvae,
the mortality of the pupae, and so on. To be informed about the effect
of the separate components, the correlation between the egg densities
of two successive generations is first studied, moving from the first to
the last generation for which data are available. Secondly, to learn the
effect of egg mortality, the correlation between the density of first instar
larvae and the egg density of the next generation is studied, and so on.
The quantitative formulatiojl of this method starts from the function
Et+i = Et . sl. S, . . . . . . s I 2 . F
where Et = egg density of generation t
= fraction surviving in the successive age intervals (s, = 1-p,, where
= fecundity (the maximum fecundity in this study, being 216 egg;gs/
8,
F
pi is the fraction apparent mortality in the relevant interval)
female )