Generation cycles, population cycles with a period of approximately one generation, have been observed in a variety of field and laboratory studies. Such dynamics are predicted to arise through the effects of resource competition and cannibalism or involve consumer-natural enemy interactions. We first show, using a new highly simplified model, that generation cycles are a very common outcome of strongly age-structured intraspecific interactions involving cannibalism. We then analyse a series of unique long-term time-series of ladybeetle (Coccinellidae) abundances from tropical Indonesia. Some of the time-series display clear generation cycles, and we argue that there is strong evidence that these are caused by intraspecific cannibalism.

To breed parasitoids, eggs, larvae and pupae of the banana skipper Erionota thrax were regularly sampled from banana fields at 7 localities in the Province of Sumatera Barat, Indonesia from November 1990 to December 1992. The sampling sites covered a wide spectrum of elevations up from the sea level and also of rainfall conditions. Twelve primary parasitoid wasps were recorded, belonging to 6 families, Braconidae (1 sp.), Chalcididae (2 spp.), Encyrtidae (2 spp.), Eulophidae (2 spp.), Pteromalidae (1 sp.) and Ichneumonidae (4 spp.), and 6 hyperparasitoid wasps, Braconidae (1 sp.), Chalcididae (1 sp.), Eurytomidae (2 spp.), Ichneumonidae (1 sp.) and Eulophidae (1 sp.). Of these wasps, a chalcid, Brachymeria lasus was both primary and hyperparasitic. Four parasitoid fly species were bred from E. thrax larvae and pupae. These belong to the Tachinidae (2 spp.), Sarcophagidae and Phoridae (each 1 sp.). Documentation is provided for each parasitoid species of (1) its distribution and abundance for each sampling site, (2) the developmental stage of E. thrax on which it oviposited and from which it emerged, and (3) the number of individuals which emerged from a single host. Local differences in the parasitoid fauna in different sampling sites are compared.

Since 1990 we have studied population dynamics of the phytophagous lady beetles, Epilachna vigintioctopunctata, E. enneasticta and Epilachna sp. 3 (aff. emarginata), (Coleoptera: Epilachninae) in Padang, Sukarami (West Sumatra), Bogor (West Java) and Purwodadi (East Java), Indonesia. These sites have distinctly different climate conditions, e.g. Padang has a typical tropical rainforest climate without a clear alternation of wet and dry seasons, while Purwodadi has a strong dry season of 6-7 months. We have conducted censuses at 3-7 day intervals, including mark-recapture of beetles and construction of life tables for up to 9 consecutive years in order to clarify the seasonal change in abundance and mortality of the populations. In most populations the adult number changed greatly with the formation of high peaks from time to time at intervals of 6-12 months. Their increase or decrease was gradual, progressing over 3-5 months. Even under conditions with sufficient amount of rainfall in Padang, Sukarami and Bogor, oviposition intensity, number of the following immature stages and, as a result, adult emergence frequently showed discrete peaks at a fixed interval. These "generation cycles" were especially clear during the phase of population increase. There was no simple relationship between the seasonal change in rainfall and that of the beetle populations. We also discussed the impacts of the strong droughts in 1994 and 1997-1998 caused by EI Niiio, and the degree of synchronization in the trends of the populations among the adjacent study sites.

Pitfall sampling of terrestrial arthropods was carried out at an arable field in West Sumatran highland (all. 900-1000 m a.s.l.) during

1. The long-term dynamics of two beetles ( Epilachna vigintioctopunctata and Epilachna enneasticta ) from Indonesia are analysed using descriptive time-series statistics and mechanistic time-series models.2. Statistical analysis of the time series reveals that the dynamics of E. vigintioctopunctata are underpinned by a second order density-dependent [AR(2)] process, which is consistent between sites. Similarly, the dynamics of E. enneasticta are also described by a second order density-dependent [AR(2)] process. The predicted dynamics from these statistical models are stable dynamics.3. Mechanistic models of the density dependence predict that in the absence of noise the dynamics of the populations should also be stable. Inclusion of the stochasticity in the models of density dependence allows a more accurate description of the population dynamics of each species to be determined.4. Correlation analyses reveal that the fluctuations between species are a function of both the density dependence and the environmental conditions experienced by each species. Under low environmental variance it is predicted that the dynamics could be synchronised but under high variance in environmental noise, the dynamics may not be synchronised and species respond idiosyncratically to the environment.5. The study highlights the value of long-term population dynamic time series for exploring the structure of population regulation and fluctuations in insect populations. Entomology, 32, 28-37 y ¼ a 0 1 a 1 Á x h1 t 1 a 2 Á x h2 t21 30 Michael B. Bonsall, Nusyirwan Hasan and Koji Nakamura