Ecological Entomology (1997) 22, 121–123
S H O R T C O M M U N I C AT I O N
Feeding by lepidopteran larvae is dangerous
E . A . B E R N AY S
Department of Entomology and Centre for Insect Science, University of Arizona, U.S.A.
Key words. Predation, feeding, Manduca sexta, Uresiphita reversalis, caterpillars, wasps, bugs.
Introduction It is difﬁcult to obtain directmeasures of predation on small animals such as insect herbivores. Instead, disappearance has sometimes been used as a rough measure of predation, or, more reliably, populations of insects inside and outside exclosures may be compared over time. In this way, many studies have deduced the importance of predation as a mortality factor for insect herbivores such as lepidopteran larvae. For example, Feenyet al. (1985) showed that predation was the largest source of mortality of Papilio polyxenes larvae. Invertebrates were most important for smaller larvae while vertebrates became important for larger ones. Much can also be deduced after natural selection has operated, about the evolutionary importance of predators by the avoidance strategies employed by caterpillars (Heinrich, 1993) Howeverimportant predation of lepidopterans may be, relatively few people have spent time observing larvae for long enough to see many predation events, except where populations of both herbivores and predators were unusually large. Examples of data that demonstrate the danger of movement include the ﬁnding that movement increased the risk of ant predation on a geometrid caterpillar (Bergelson & Lawton,1988). Similarly, caterpillar movement was shown to increase predation by a pentatomid bug (Marston et al., 1978) and a crab spider (den Boer, 1971). So, while it may be assumed that movement is generally dangerous, less is known about the risk of actually feeding. This study is a quantiﬁcation of the risk of predation by two predators, posed by feeding in two caterpillar species. Methods and ResultsAll data were obtained by continuous observation of individual caterpillars of two species, over periods lasting at least 6 h. One study involved early larval stages of the pyralid, Uresiphita reversalis, feeding on its most common host in northern California, Genista monspessulanus (broom). The second study involved the sphingid, Manduca sexta (tobacco hornworm), feeding on cultivated plants ofNicotiana tabacum (tobacco).
Uresiphita reversalis Continuous observations were made on natural populations near Berkeley, California. The caterpillars were mostly ﬁrst and second instars occurring in groups near the tips of branches. Observations lasted from dawn to dusk, approximately 12 h. Each of 298 caterpillars was recorded resting, moving or feeding. A total of 3800 ‘caterpillar hours’ ofobservations was made. One or two groups of individuals could be observed closely at one time from a chair taken into the ﬁeld and placed beside the shrub. Although U. reversalis is aposematic (Bernays & Montllor, 1989) and distasteful to ants and wasps (Montllor et al., 1991), other invertebrate predators kill up to 20% of larvae in the ﬁrst two or three instars. At this stage in the lifehistory, the larvae are gregarious and spin loose silken shelters between leaves or even over a single leaf. After hatching from the eggs, which are deposited in groups of variable size, the young larvae can feed for a day or so within the shelter they make. Here, they skeletonize the leaves. After that time, they enlarge the shelter somewhat, but individuals must make regular forays to feed on otherleaves. The forays consist of movements of several centimetres, a feeding bout of several minutes, after which they return to the shelter. In the shelter they rest and moult. During the period of this study, the major predator was a species of anthocorid bug that resided on the broom plants and presumably fed upon various small arthropods and their eggs. Occasional predation events were observed...