On the stabilization of animal numbers. Problems of testing : 3. What do we conclude from significant test results?

When testing census data of insect populations for regulation, and/or for overall density dependence in the course of numbers over years, certain conditions, which follow from the testing models, should be fulfilled. Even if the series of densities may be considered a piece of first-order Markov chain (a necessary condition) significant test results need not obviously point to regulation of numbers by dominant density-dependent processes. Such a case is presented by the pine looper population at "Hoge Veluwe" studied by Klomp. A drastic drop in density from 1952 to 1953, which takes 78-97% of the log-density range (LR) over all years, most probably wrongly causes significant test results. This is supported by some simulation experiments. Moreover, we cannot be sure that the population was sufficiently isolated, i.e. that dispersal of adults from surrounding populations did not importantly influence population numbers. Among 6 Panolis-populations studied by Schwerdtfeger during 17 years a single one scored significantly with all tests. This resulted, however, from such a drastic drop in density that it covered the entire log-density range (LR=9.39), which therefore is wider than in any of the other (non-significant) populations. Another Panolis-population that maintained itself during 60 years, and which also scored significantly, most probably was kept within limits by supplementation of very low densities with immigrants, on the one hand, and by restriction of high densities by defoliation caused by other species, on the other. It is discussed whether this can be considered "regulation", or results from spreading of risk. It is concluded that the range stability of particular populations must be considered generally to be the result of stabilization by both internal and external processes among which both density-dependent and density-independent processes play a significant part, and from which the contribution of the density-dependent processes need not be separated. The most interesting aspect of the stabilization of animal numbers is its relationship with the expected survival time of the population.