3. It is not known what unnatural effects the glass windows of the rhizotron introduce into the system. They produce a solid surface that is an obstacle for animals and plant root movement; the temperature could be effected and there will be some light effects, if only from the electronic flash of the time lapse camera. The substantial clay content of the soil (Sackville-Hamilton, et al. 1991) means that it cracks in drought conditions. These cracks always occur between the window and the glass, so opening fissures down which animals can move. This addressed in paper 5 where it is seen to be a common event even when no glass is involved. However, attempts were made to minimize soil retraction by irrigation.
4. As the investments in the rhizotron is substantial, it is not feasible to replicate it in different soil types or within the soil type.
Nevertheless, we can begin to record and measure the incidence of predation underground, in a way which has previously not been thought possible (McBrayer and Reihle, 1971). More especially, rhizotron observations extend the observations on surface active predators and add much new information on the activity of true soil invertebrates such as geophylid centipedes and diplurans.
Even when artificially placed prey are used, contact with potential predators occurs infrequently and recording these events at one speed, but viewing at another so that events are greatly speeded up (time lapse), allows rapid scanning, followed by concentration on those periods when activity is occurring. Lussenhop et al. (1990) observed soil invertebrate activity using a video camera, but the continous lights required by the video may disrupt predator activity and increase the temperature of the soil. Time-lapse cinematography has the advantage of using electronic flash, which circumvents these problems. However it is not without disadvantages, and selecting the time-lapse interval (8 minutes in the present study) is a compromise between saving time and resources and the increasing likehood of missing key events. In the present study, contacts between prey and predators are recorded, and continous feeding is readily observed, but the actual act of killing the prey is much less frequently seen.
Not all resting stages are equally vulnerable to predator attack in the soil, and the pupae of T. molitor were selected as a prey in the present study, because they are more vulnerable than the eggs tested. Stick insect eggs appeared to be well protected by their hard shells, and the emerging nymphs would escape from the soil, whilst mustard beetle pupae were small, and again many emerged before being attacked. These must represent some of the ways in which resting stages escape predation, but these defence mechanisms were not explicitly studied in the present thesis.
The number of contacts with potential prey is an easily measured parameter used extensively in the present study. It is a measure which reflects both the population of predators and their activity, in a similar way to catches in a pitfall trap. However unlike pitfall traps, where individuals are caught only once, the number of contacts per prey often reflects repeated contact by the same predator as it investigates and then begins to attack and feed on the pupa. To this extent, contacts are not independent events and this introduces statistical problems. However, in the present study, it is the number of contacts which is of interest, as it reflects the level of danger to which the prey is exposed and this measure carries no important implications about the number of predators involved. The technical effect is to produce aggregation in the counts for the number of contacts.
The present study has shown that typical surface active predators such as carabids and staphylinids which are encountered frequently in pitfall trap studies, also hunt below ground. In the case of laboratory studies, it is clear that they can burrow through loose soil and retrieve pupae from at least 6 cm. However, this energetically expensive hunting method is used much more frequently when the more readily available prey on the surface have been consumed. In the rhizotron, it was shown that supplying artificial tunnels from the surface to the lower layers where pupae were placed, greatly enhanced their rate of disappearance. There was also evidence that in dry conditions when the soil began to contract back from the glass, predators foraged down these cracks, and this can sometimes be seen in the films. The studies on the extent of cracking in the clayey soils of Treborth show that this is a natural and extensive phenomenon in dry weather, and must provide a substantial extension of the foraging range for these normally surface active animals (see paper 5). Some groups such as carabids seem to penetrate deeper, and play a more important role than others such as staphylinids.
What the present study has shown is the important role of the true underground fauna, which are not normally sampled in pitfall traps. Diplurans are a dominant group in this area. Although it is less clear how often they are responsible for killing a healthy pupa, there is no doubt of their importance in exploiting the pupae once they have died. Similar comment can be made about the roles of slugs and earthworms. Geophilid centipedes are specifically adapted to an underground predatory life style, where they use existing tunnels, often earthworm tunnels to move about. This has important implications for resting stages, which will be at increased danger of attack if they use existing worm burrows, or if a worm contacts them during the process of burrowing.
The ways in which underground predators interact is important but not understood. The intriguing observations in paper 5 suggest that as the surface predators penetrate further underground once resources on the surface become exchaused, the obligatory underground feeders such as Campodea go deeper. It may be that this is a response to increased competition for food, or it may be that predators such as carabids and staphylinids will also attack Diplura, which take avoiding action. In either event, the very deepest prey will come under increased pressure.
In general, the data collected in this study, both for the number of contacts with predators and the frequency of pupal disappearance, all suggest that it is safer for resting stages to be underground rather than on the surface and the deeper the animal goes the safer it becomes. Moreover, the diversity of predators is reduced with depth. Predation above ground by invertebrate predators is markedly relaxed during the winter, as the cold conditions, for which the overwintering inactive forms are an adaption, will also reduce the activity of predators. Vertebrate predators by contrast are likely to enhance their predation on these forms, as they must continue to find food over the winter to maintain body temperature, and there are few active invertebrates to hunt. At this time af the year, birds, shrews and rodents may be important surface active predators. By contrast, we have seen in Chapter 6 that invertebrate predators remain active below ground throughout the winter. Vertebrates as underground predators are restricted by the high energy costs of traveling. Moles specialize on large energy-rich and abundant lumbricid fauna could not survive on insect resting stages. Birds rely on bill length to probe into the soil, rather than on digging, and this is most effective in wet muddy conditions. The fact that so many insects do overwinter in the ground (Table 1) is a measure of the protection they must receive from adverse climatic conditions, vertebrate predators, and the difficulty of being located at depth even by specialized subterranean predators.
The rates at which Tenebrio pupae disappeared even at depth in the present study, should not however be taken to represent normal rates of disappearance. As stated, these pupae have not been selected for a subterranean life style (Beddington, 1975); they are present in much greater densities, and at much higher levels of aggregation that would normally be the case, and the earth had to be disturbed to place them there. All these factors may enchance rates of loss (Hassels & May, 1974; Hassel et al. 1976; Hassel et al. 1977; Hassel, 1985; Hassel & May, 1986 and Hassel, 1987).
