Time tells: long‐term patterns in the population dynamics of the yew gall midge, Taxomyia taxi (Cecidomyiidae), over 35 years

Abstract.  1. Three populations of the yew gall midge, Taxomyia taxi , and its two chalcid parasitoids, Mesopolobus diffinis and Torymus nigritarsus , from Kingley Vale, West Sussex, U.K., were monitored each year between 1966 and 2002. Taxomyia taxi 's life cycles are either hemivoltine or univoltine and the parasitoids' are univoltine ( T. nigritarsus ) or multivoltine ( M. diffinis ). 2. Growth of the host trees, yew Taxus baccata , each year between 1950 and 1997 was assessed by width of annual rings in cores taken in 1998. Temperature and rainfall data were provided by the weather station in Southampton, representative of the climate of central southern England. 3. Time‐series analyses were conducted to explore the effects of climate, tree growth, and parasitoids on each T. taxi population and on the three populations combined. Four time‐delayed models were developed: for hemivoltine and univoltine T. taxi and for each of the parasitoids. 4. Hemivoltine T. taxi populations showed troughs in density in the periods 1972–1975 and 1991–1992, which were more pronounced in even than odd years; these were closely tracked by M. diffinis and followed, after a delay, by T. nigritarsus . Time‐series models showed that, for T. taxi , its current density depended on the density of its hemivoltine parents moderated by the negative effects of T. nigritarsus on these parents. In reciprocal fashion, T. nigritarsus density was influenced by that of its hemivoltine host as well as by its own previous density. Host and parasitoid appear to participate in a coupled interaction that regulates the populations of both via delayed density dependence, resulting in cycles of at least 14 years. Mesopolobus diffinis had no effect on the dynamics of hemivoltine T. taxi , most of which escaped its parasitism. Although univoltine T. taxi can give rise to hemivoltine offspring, there was no evidence that they affected hemivoltine dynamics. 5. The density of univoltine T. taxi was usually much lower than hemivoltine density and varied irregularly over time. It was not attacked by T. nigritarsus but parasitism by M. diffinis was always high. Despite this, its density depended only on the density of its parents, especially the hemivoltine parents. Parasitism had no effect on the dynamics of univoltine T. taxi . Conversely, M. diffinis density depended solely on the densities of both univoltine and hemivoltine hosts. 6. Tree growth appeared to have little effect on the dynamics of the host or parasitoids and there was no evidence that gall density had any effect on growth of the tree. The primary effect of climate appeared to be in maintaining the cyclic nature of the T. taxi–T. nigritarsus interaction. If stochastic weather variables are removed from the models, the oscillations are of equivalent periodicity but dampen to equilibrium.