SCIENTIFIC CORRESPONDENCE

moving within a forest stand (Fig. la); Even smaller radar tags on after 48 hours, the moth stage of their tent caterpillar hosts was relocated up to 100 m SIR - Movement behaviour is funda­ from the point of release (Fig. lb). In a mental to the understanding of mark-recapture study of the Apollo butter­ population dynamics 1• Unlike researchers , we repeatedly recaptured 10 individuals working on vertebrate species, insect ecol­ fitted with tags, together with untagged ogists have not had the benefit of such , over a period of 2 weeks. But­ technology as radio-collaring for tracking terflies did not seem to be affected by the movements. Walking beetles have 0.4-mgtags. been fitted with 'tags' that reflect harmon­ Movement data can be collected by one ic radar2; these tags have now been or more researchers, each carrying a reduced in size so that bees can carry transmitter-receiver. Use of tags small them and be detected using powerful enough so as not to alter insect move­ radar signals3• To ensure that the tag ment, in combination with more powerful attached to the insects does not itself alter transmitters, may provide a tool for study­ their normal movement, its size and ing movement, the 'missing element' of weight must be extremely small. Here we insect population dynamics. report extreme miniaturization of such a Jens Roland tag to an order of magnitude smaller than Department of Biological Sciences, that previously used°, and describe its use University of Alberta, Edmonton, with portable radar equipment in the field Alberta T6G 2E9, Canada on a variety of insect species (Fig. 1). Graham McKinnon The tag (Fig. 2) is glued to the insect Chris Backhouse thorax or abdomen with rubber cement, Alberta Microelectronic Centre, and the near-invisible dipole simply trails University of Alberta, Edmonton, behind the insect. This configuration is Alberta T6G 2T9, Canada two orders of magnitude lighter than that Philip D. Taylor used for tracking ground beetles2, and an Department of Biology, Acadia University, order of magnitude lighter than that used Wolfville, Nova Scotia BOP 1XO, Canada for tracking bees3. Our transmitter- receiv­ 1. Turchin, P. Ecology 72, 1253-1266 (1991). FIG . 1 Relocations and trajectories from a er (RECCO Rescue Systems, Lidingo, 2. Mascanzoni, D. & Wallin, H. Ecol. Ent. 11, 387- 390 release point for: a, a single Patelfoa pachy• (1986). Sweden) is carried in a portable backpack, pyga fly over a 3-h period; b, c, forest tent 3. Riley J. R. et al. Nature 379, 29-30 (1996). weighs 8 kg and transmits a 1.7-W contin­ caterpillar moths over 48 h, all within forest uous microwave frequency of 917 MHz stands of trembling aspen (shaded) adjacent from a hand-held Yagi antenna. The tag to clearings (unshaded). Reducing antibiotic on the target insect reflects this energy at a frequency of 1,834 MHz, which is resistance detected by the same hand-held antenna. The use of a longer fre­ SIR - A reduction in the use of antibiotics quency than that used for tracking has recently been proposed as a measure to bees3 reduces the attenuation of forestall, and ideally reverse, the growing the signal by water in intervening public-healt~ Eroble°:1 of antibiotic-re~is­ vegetation. The current configura­ tant bactena -3_ This recommendation tion of transmitter- receiver and implicitly assumes that resistance to anti­ dipole permits detection at up to 50 biotics reduces the fitness of bacteria in the m. Tagged insects are relocated at absence of antibiotic-mediated selection, intervals by systematic searching thereby allowing sensitive bacteria to and triangulation on the reflected replace resistant strains in drug-free signal. environments. We now show that, although We have successfully fitted tags to mutations to antibiotic resistance in bac­ five species of flying insects with a teria may initially impose a considerable range of body weights: the fitness burden, natural selection can result small Apollo Pamassius in the rapid fixation of second-site, com­ sminthius (350 mg), the common pensatory mutations which substantially alpine butterfly epipsodea reduce this cost by restoring physiological (150 mg), both the larvae and moths functions impaired by resistance mutations of the forest tent caterpillar Malaco­ without altering the level of bacterial soma disstria (200 mg; Fig. 2a ), and resistance. the parasitic tachinid fly Some point mutations in the rpsL gene pachypyga ( 45- 60 mg; Fig. 2b) and in the bacterium Escherichia coli confer sarcophagid fly Arachnidomyia resistance to high concentrations of strep­ aldrichi (55- 75 mg), both of which tomycin, but reduce the darwinian fitness attack the tent caterpillar. The tag of bacteria by decreasing peptide chain increases the weight of these insects elongation rates4•5• We considered two FIG. 2 Forest tent caterpil lar moth (a) and tachinid fly by only 0.1-0.9%. Tagged insects (b) fitted with tags made from a spontaneous streptomycin-resistant (Str'), can be used to examine the effect of Schottky low-barrier diode (Hewlett-Packard, HP 5082- rpsL mutants of E. coli CAB281 (see habitat structure on insect move­ 2837) mounted in the centre of an 8-cm dipole anten• table). One mutant had a 14% per genera­ ment. Over a period of 3 hours, we na of extremely fine aluminium bonding wire (arrows). tion disadvantage relative to the Str' repeatedly located tagged parasitic Total weight of this configuration is 0.4 mg. parental strain, and the other had a 19% 120 NATURE · VOL 381 · 9 MAY 1996