Dragonfly larvae catch prey by rapid protraction of a curiously modified labium. There are two main types of predatory behavior; 'climbers' move amongst aquatic vegetation and use their large compound eyes to detect prey at a distance and subsequently to orient towards it, while 'sprawlers' live at the bottom and detect most of their prey by tactile stimulation. Five species of dragonfly larvae were successful in 60 to 70% of their attacks on mosquito larvae, but in only 25 to 28% on gammarids. Size and movement of prey are important in its recognition, while shape, color, and odor are not. The forward thrust of the labium takes only 15 to 20 milliseconds, and at the time of the strike the anal siphon is closed so that the blood pressure, which provides the force for labial extension, is all directed forwards and is not used to expel water from the colon.

Average reaction norms relating number of day-degrees required to complete egg development to temperature are described for 95 species (1 15 populations) of Plecoptera, Odonata, Ephemeroptera, and Diptera (mainly Culicidae, with some Tipulidae and Sciomyzidae). The slope of the average reaction norm is used as an index of adaptation, with positive slopes indicating cold-adapted species, negative slopes indicating warm-adapted species, and slopes around zero indicating generalist species; 57% of the between-taxon variation in slope is associated with differences between orders, 14% among families, 7% among genera, 14% among species, and 8% among populations. Significant differences between congeneric species are found in only 4 of 92 possible comparisons. Only Plecoptera show much coldadaptation, with 40% of the species having significantly positive slope. However, 26% of the species (mainly in the Systellognatha) have significantly negative slope, suggesting adaptation to warmer waters than those in which the order is believed to have originated. The other orders probably originated in warm water and have generally maintained this adaptation. All Odonata, 7 1 O h of the Diptera and 81 % of the Ephemeroptera have significantly negative slopes. Diapause is a common alternative to metabolic adaptation to deal with unfavourable thermal environments. We suggest that it occurs widely in eggs of Plecoptera and in the mayfly Rhihgcna h y o b , cold-adapted species that may use diapause to survive periods of high temperature.

The natural food of pond-dwelling larvae has been studied by analyzing the contents of faecal pellets and comparisons are made among different species, different sizes of larvae, different habitats, and different times of year. Dragonfly larvae are general animal feeders and the composition of their food in nature is affected by a complex of factors including the relative abundance of different prey in the environment, the size and habits of this prey, and the ease with which it is caught and devoured.

We separate the mating systems of odonates into two main groups: non-resource and resource-based systems. These two groups comprise five classes of mating system: encounter-limited mating, free female choice, resource-limitation, resource-control and female-control. These classes are consistent with previous classifications of odonate mating systems and with the overall classification of mating systems by Emlen & Oring (1977: Science, 197: 215-223). Whereas Emlen & Oring's classification of mating systems was concerned with differences in sexual selection between mating systems, our classification of odonate mating systems also addresses the influence of inter-and intra-sexual selection on males within a mating system. Predictions about such relationships are useful in multivariate analysis of odonate lifetime reproduction success. Among most odonate mating systems, much of the sexual selection on males results from male-male competition for access to mates. Sexual selection via female choice is relatively less important or operates indirectly through females' choices of times or places to mate. We place resource-control and resource-limitation at opposite ends of a resource-defence continuum and postulate female choice will have greater influence in mating systems that are more like a resource-limitation system and less influence in mating systems that are more like resource-control. Sexual selection is likely to be weak in species that resort to encounter-limited mating where longevity is likely to contribute strongly to variation in reproductive success. Females have limited opportunity to exercise choice among males in the female-control mating system and in this system selection is most likely to operate on male characters which contribute to their efficiency in searching for and capturing mates. Predictions about the differences in the intensity of sexual selection between different odonate mating systems should be made on the basis of the variation in the number of potential fertilizations per male or even per ejaculate, rather than the number of fertilizable females per male. Very different mating systems could result in similar patterns of variation in male reproductive success.

No abstract