12. CCP4: A Suite of Programs for Protein Crystallog- raphy (SERC Daresbury Laboratory, Warrington, UK, 1986). 13. K. D.Cowtan and P. Main, Acta Crystallogr. D49, 148 (1993). 14. G. J. Kleywegt and T. A. Jones, in preparation. 15. A partial model comprising 461 of the 517 resi- dues was built in the program 0 [T. A. Jones and M. Kjeldgaard, 0-The Manual (Uppsala Univer- sity, Sweden, 1991)l. This model was refined in the program X-PLOR [A. T. Brijnger, X-PLOR Ver- sion 2.1 (Yale University, New Haven, CT, 1990)]. Calculated and model-derived phases were com- bined in the program SIGMA4 [R. J. Read, Acta Crystallogr. A 42, 140 (1986)],allowing the re- maining residues to be added. Subsequently, the structure was refined independently for the two crystal forms with the program X-PLOR; manual model adjustment was done in the program 0; and cycles of least squares refinement were car- ried out with the program PROLSQ (12). The 4550 Fig. 4. Stereo diagram showing potential hydrogen bonding and electrostatic interactions (dashed OppA-U structure, which comprises atoms, including 354 water molecules, two uranium at- lines) between the trilysine ligand (thicker bonds) and OppA. oms, and the tetrapeptide ligand Val-Lys-Pro-Gly, has a crystallographic R factor of 16.7% for all data (29,634 reflections) between 8 and 2.1 A gand binding appears, therefore, to be based 241,161 (1973);S. M. Hammond et ab, ibid. 327, spacing and root-mean-square (rms) deviations 730 (1 987). of bond lengths and angles from ideal values of chiefly on the avoidance of potentially un- 5. M. S. Hanson et ab, Infect. Immun. 60, 2257 0.015 A and 0.037 A, respectively. The OppA-L favorable interactions with the repertoire of (1992);W. N. Abouhamad etal., Mol. Microbiol. 5, structure, which was refined against data collect- ligand side chains. 1035 (1991); C. Navarro et a/., ibid. 9, 1181 ed at -15O0C,comprises 4663 atoms, including (1993); E. Olson et a/., J. Bacteriol. 173, 234 432 water molecules, eight uranium atoms, five The burial of peptide ligands within (1991); R. Tan and M. H. Saier, Microbiol. Rev. 57, acetate ions, and the tripeptide Lys-Lys-Lys, and OppA according to the Venus flytrap mech- 320 (1 993). has an R factor of 17.3% for all data (47,801 10 1.8 anism in some sense represents the final 6. J. Payne and C. Gilvarg, Adv. Enzymol. 35, 187 reflections) between and A spacing. The (1971). rms deviations in bond lengths and angles from stage of a folding process to form a protein 0.018 0.034 7. J. 329, 512 (1987); ideality are A and A, respectively. with a variable peptide core and a unique P. Bjorkman et ab, Nature This work was supported by the Medical Re- D. R. Madden et a/., ibid. 353, 321 (1991); M. J. surface structure that is efficiently recog- 262,1234 (1 993). search Council, the Science and Engineering Kuehn et a/.,Science Research Council, and the Imperial Cancer Re- nized bv the membrane comoonents of the '8. C. F. Higgins, Nature 327,655 (1987). search Fund. C.F.H,is a Howard Hughes Interna- transport system. The manner in which 9. P. J. Kraulis, J. Appl. Crystallogr. 24,946 (1991). tional Research Scholar. We would like to thank l. OppA accommodates the diverse peptide 10. S. P. Tolley etal., J. Mol. Biol. 204,493 (1988); 1. Glover, R. Denny, and P. Moody for assistance in D.Hiles and C. F. Higgins,- - Eur. J. Biochem. 158, collectina x-ray data and D. Edwards, U. Gileadi, side chain functional groups is quite distinct 561 (1986). S. ~~de,-andS. Mowbray for helpful discussions. from the way this is achieved in major 11. W. A. Hendrickson, Science 254, 51 (1991); I. D. histocompatibility complex molecules and Glover et a/,,in preparation. 22 February 1994;accepted 25 April 1994 chaperone proteins. In the crystal structures of the latter, the peptide ligands are located on the surfaces of the molecules and many Habitat Fragmentation, Species Loss, and of the side chains are directed into the - solvent (7). Biological -control The covalent coupling of synthetic an- tibacterial compounds to peptides has Andreas Kruess* and Teja Tscharntke proved a successful route to overcoming the problem of membrane impermeability Fragmentation of habitats in the agricultural landscape is a major threat to biological assdciated with some of these drugs (4). diversity, which is greatly determined by insects. Isolation of habitat fragments resulted in This "Trojan horse" approach (8) relies on decreased numbers of species as well as reduced effects of natural enemies. Manually the unusually broad specificity of the oli- established islands of red clover were colonized by most available herbivore species but gopeptide permease. Knowledge of the few parasitoid species. Thus, herbivores were greatly released from parasitism, experi- structure of the initial receptor for this encing only 19 to 60 percent of the parasitism of nonisolated populations. Species failing transport system, OppA, will be valuable to successfully colonize isolated islands were characterized by small and highly variable in guiding the design of effective peptide- populations. Accordingly, lack of habitat connectivity released insects from predator based antibiotics. control. REFERENCESANDNOTES 1. G. F.-L. Ames. Annu. Rev. Biochem. 55. 397 Research in conservation biology analyzes ly results in changes in community structure (1986); C. F. ~i~~ins,Annu. Rev. Cell Biol.'8, 67 (1 992). both species richness and the functioning or and function, including loss of species in 2. F. A. Quiocho, Curr. Opin. Struct. Biol. 1, 922 stability of ecosystems. Modern agricultural isolated islands and disruption of the food (1991),and references therein; B.-H. Oh et a/.,J. methods fragment natural ecosystems; and web (I). Communities of herbivorous in- Biol. Chem. 268, 11348 (1993); S. L. Mowbray and L. B. Cole, J. Mol. Biol. 225,155 (1992). the subsequent increase in isolation typical- sects and their natural enemies (such as 3. J. Payne and C. Gilvarg, J. Biol. Chem. 243,6292 parasitoids) centered on a single plant spe- A. Kruess, Zoologisches lnstitut I, Universitat, Korn- (1968);C. A. Guyer et ab, J. Bacteriol. 168, 775 provide a small in which the blumenstrasse 13. D-76128 Karlsruhe. Germanv. cies ecosystem (1986);E. W.Goodell and C. F. Higgins, ibid. 169, interactions between the organisms can be 3861 (1987). T. Tscharntke, FG Agrar6kologie, Universitat, ~ald- weg 26, D-37073 Gottingen, Germany. 4. B. N.Ames etal., Proc. Natl. Acad. Sci U.S.A.70, experimentally analyzed. Insects amount to 456 (1973); T. E. Fickel and C. Gilvarg, Nature *To whom correspondence should be addressed. more than half of all living organisms and SCIENCE VOL. 264 10 JUNE 1994 1581 have a greater impact on terrestrial ecosys- feeding weevils (A. apricans and A. assimile) ing conclusions from experimental predator tems than any other type of animal (2, 3). always have large wings and are thereby removal (8). For the stem-boring Apion spp., For this report we tested the hypothesis more able to follow changing resources (6). parasitism in isolated clover patches was only that extinctions in small and isolated hab- Such increased flight capability may carry a 30% of that found in larger ecosystems (Fig. itat patches should not affect all insect cost of reduced fecundity but enables high 2B). Host mortality in the five most isolated species equally. Natural enemies of phy- dispersal rates and success in the unpredict- patches was only 19% of that observed in tophagous insects are expected to become able, fragmented environment of the agri- control patches. For the seed-feeding Apion extinct first, thereby increasing the risk of cultural landscape (7). spp., average parasitism in clover islands was pest outbreaks (3, 4). We analyzed commu- Numbers of parasitoid species attacking 75% of that found in the controls (Fig. 2C), nities of stem borers and seed feeders of red the clover herbivores were neeativelvu cor- because the responsible parasitoid, S. dubius, clover (Trifolium pratense). Eighteen clover related with distance between clover islands was very abundant in all 18 clover patches. habitats, each covering 1.2 mZ, were ex- and meadows. Eieht- to twelve ~arasitoid The seed feeders in the five most isolated perimentally established in an agricultural species were found in the meadows, but islands had 60% of the control parasitism. landscape dominated by crop fields and only two to four species in the patches 500 Spintherus dubius is the most abundant and small, fragmented meadows (northeast of m from the nearest meadow (Fig. 2A). widespread chalcid wasp in set-aside fields and Karlsruhe, Germany). The clover islands Numbers of species were closely correlated fallows of the study area (9). If S. dubius was were separated from the nearest meadow with both numbers of parasitoid individuals excluded from the analysis (Fig. 2D), parasit- with .naturally occurring clover plants, by (regression analysis, F = 23.7, r2 = 0.597, ization decreased to 10% on the most isolated 100 to 500 m, so that effects of habitat n = 18, P < 0.001) and percent parasitism patches. connectivity could be tested. (F = 21.2, rZ= 0.570,n = 18, P < 0.001). Therefore, habitat isolation affects spe- The endophage insect community was Therefore, isolation of the clover islands cies diversity, having the greatest impact on identified by dissection and rearing (Tables reduced both parasitoid diversity and the parasitoids, thereby releasing pest insects 1 and 2). Of the eight herbivore species, percent parasitism (Fig. 2, B to D), support- from parasitism and potentially allowing five fed on seeds in the flowerheads (Apion apricans, A. assimile, A.