ANALYSIS O F PROCESSES I N CONTROL O F INSECTS
17
Only in two cases in ten do observation and expectation disagree.
Klomp writes: “Under the hypothesis that no tendency to return to
the constant level is operating, we can expect that in half of the cases
there is no agreement. The probability of the above result under this
hypothesis is 0-03 (tested one-sided). Consequently, this hypothesis can
be rejected.”
This gratifyingly simple method can succeed only with highly
regulated populations, where the regulatory action is dominant over
the non-regulatory fluctuations. Where the opposite is the case, the
method does not detect the regulatory element. This can be demonstrated by means of the population graph J in Fig. 1 0 ~ .
This represents an imaginary population in which the main fluctuations are
density-independent, but in which the rate of change upwards or downwards is density-dependent. The scale of density used is logarithmic,
and the mean is the geometric mean, 2.26. The graph provides thirteen
steps in which the expected and the observed trends can be compared.
There are only six agreements, and seven disagreements.
b. After arti$cially imposed changes. Nicholson (1957) and Hairston
(1957) have remarked that it should be possible to demonstrate the
existence of regulation artificially by imposing a change in density
upon a field population and observing its returns towards the normal
level. Such experiments would have to be done with care to avoid
interference with natural enemies, where these are important ; for this
reason, many pest control operations would not provide a suitable test;
I
I
FIG. 7. Simplified model in which a population tending to assume a constant density is
reduced to lower densities by exploitation (vertical arrows), and its subsequent increase
observed. (From Solomon, 1962b.)
17
Only in two cases in ten do observation and expectation disagree.
Klomp writes: “Under the hypothesis that no tendency to return to
the constant level is operating, we can expect that in half of the cases
there is no agreement. The probability of the above result under this
hypothesis is 0-03 (tested one-sided). Consequently, this hypothesis can
be rejected.”
This gratifyingly simple method can succeed only with highly
regulated populations, where the regulatory action is dominant over
the non-regulatory fluctuations. Where the opposite is the case, the
method does not detect the regulatory element. This can be demonstrated by means of the population graph J in Fig. 1 0 ~ .
This represents an imaginary population in which the main fluctuations are
density-independent, but in which the rate of change upwards or downwards is density-dependent. The scale of density used is logarithmic,
and the mean is the geometric mean, 2.26. The graph provides thirteen
steps in which the expected and the observed trends can be compared.
There are only six agreements, and seven disagreements.
b. After arti$cially imposed changes. Nicholson (1957) and Hairston
(1957) have remarked that it should be possible to demonstrate the
existence of regulation artificially by imposing a change in density
upon a field population and observing its returns towards the normal
level. Such experiments would have to be done with care to avoid
interference with natural enemies, where these are important ; for this
reason, many pest control operations would not provide a suitable test;
I
I
FIG. 7. Simplified model in which a population tending to assume a constant density is
reduced to lower densities by exploitation (vertical arrows), and its subsequent increase
observed. (From Solomon, 1962b.)