SCIENTIFIC CORRESPONDENCE

from two variable portions of the mito­ Ancient DNA and island endemics chondrial control region (133 and 312 base pairs, respectively). Comparison of Sm - Most recently extinct and currently migratory ( platyrhynchos), a the sequences ( see figure) clearly shows endangered species of are inhabi­ perception that has greatly influenced the that the subfossil bones match those of tants of islands, where the effects of recovery programme of the duck7•8. the extant , indicating that the anthropogenic and habitat However, the relationships between the species was formerly widespread in the modification have been most severe1• three taxa were uncertain7•9• Recent genet­ . Recently, palaeontological records have ic studies have demonstrated that, while Late Holocene subfossils of Laysan shown that the effect of prehistoric the koloa is indeed closely related to mal­ ducks, including some bones of non-flying humans on insular biotas, particularly in lards, the Laysan duck is very distinct from juveniles, have been found in habitats the Pacific, were far more severe than pre­ either (J.R., unpublished data). varying from near sea level on the islands viously believed2-4, and that many extant Bones of small ducks occur in Late of , and to formerly flighted species that now appear to be Pleistocene and Holocene deposits in the forested areas at high elevations (60-1,800 endemic to single islands were previously chief Hawaiian islands3•10. Measurements m), far from permanent water, on more widespread4. Here we report that of the lengths of the main long bones, with and . Although the species was the combination of ancient DNA and sample sizes between 3 and 27, showed evidently widely adaptable, it is highly palaeontological techniques can provide the average lengths of the palaeontol­ unlikely that Laysan island (maximum information necessary for conservation ogical specimens to be intermediate altitude 12 m) presents an optimal habitat management of such a species, the endan­ between Laysan ducks and koloa (dns). for it. gered Laysan duck (Anas laysanensis). The bones could not be identified from The disappearance of the Laysan duck The Laysan duck is historically known these data because: (1) skeletal morphol­ everywhere but on Laysan island is only from the remote, small (370-hectare) ogy is poorly diagnostic in dabbling assumed to result from the same human­ island of Laysan in the northwestern ducks9; (2) other island ducks changed induced factors (hunting, habitat destruc­ Hawaiian chain, where its numbers have size during the Holocene 11 ; and (3) the tion, introduction of predators and varied between 20 and approximately 500 bones could represent an extinct taxon pathogens) that are blamed for the extinc­ over the past 70 years5·". Fewer than 150 distinct from either the koloa or Laysan tion of other taxa native to Pacific islands. survived drought conditions in 1993, illus­ duck. Our data justify the reintroduction of the trating the vulnerability of this population. To resolve the identity of the subfossil Laysan duck to parts of its former range in Both the Laysan duck and the bones, we used PCR (polymerase the main Hawaiian islands where adverse duck, or koloa (Anas wyvilliana ), were chain reaction) and ancient DNA tech­ factors can be controlled. This study once thought to be derived from stray niques12 to extract and sequence DNA demonstrates the value of using ancient DNA and palaeontological information to design recovery plans for endangered species and should provide a model that Laysan duck 1 can be adapted to many other threatened 100 Laysan duck 2 insular species in the Pacific and elsewhere. ------t Laysan duck 3 Alan Cooper* Subfossil 1 Judith Rhymert Subfossil 2 Helen F. Jamest Storrs L. Olsont 99 Carl E. McIntosh* Koloa 1 Michael D. Sorenson* Koloa 2 Robert C. Fleischer* Koloa 3 *Molecular Genetics Laboratory, 63 National Zoological Park, ------ 1 Smithsonian Institution, African Washington, DC 20008, USA ------black duck tDepartment of Biological Sciences, Clemson University, Clemson, South Carolina 29634, USA 0.12 0.09 0.06 0.03 0 'fDepartment of Vertebrate Zoology, Jukes-Cantor corrected distance National Museum of Natural History, Smithsonian Institution, 13 Neighbour-joining two-parameter-corrected distance tree obtained with MEGA . An identi• Washington, DC 20560, USA cal topology is produced by parsimony analysis. with a consistency index of 0 .9 (PAUP 3.1.1; ref. 14), and bootstrap values from 1,000 replications are given. The control 1. Milberg, P. & Tyrberg, T. Ecography 16, 229- 250 (1993). 2. James. H. F. & Olson. S. L. Omith. Monogr. 45, Pt II. region sequences were amplified with PCR using primers: C1 (L 00078) 5'-GTTATTTGGT• (Am. Ornith. Un., Washingt on DC, 1991). TATGCATATCGTG; C1R 2 (H 00390) 5'-CGATTAGTAAATCCATCTGGTAC; C2 (L 00736) 5'• 3. Olson, S. L. & James, H. F. Omith. Monogr. 44, Pt I (Am. ATCTAAGCCTGGACACACCTG; t-Phe (H 01251) 5'-TGGCAGCTTCAGTGCCATGC; and GCDR Ornith. Un., Washington DC, 1991). 4. Steadman, D. Science 267, 1123-1131 (1995). (L 01117) 5'-TATTAGAGAAACTCCAGTAC , where the strand designat ion and 3' position in 5. Warner, R. E. Condor 65, 3-23 ( 1963). the published chicken sequence are given in parentheses. Sequences were obtained as 6. Marshall, A. P. Bird Conserv. Int. 2 , 239-251 (1994). described12 from an African black duck outgroup, two subfossils from different caves on 7. Weller, M. W. The Island Waterfowl (Iowa State Univ. Hawaii, and three koloa, three Laysan ducks and two genetically divergent mallards. Press, Ames, 1980). 8. Moulton, D. W. & Weller, M. W. Condor 86, 10 5- 117 Considerable genetic diversity has been detected within mallard tax a (J.R., unpublished (1984). data), so the two most divergent haplotypes are used in this analysis. The subfossil 9. Livezey, B. C. Wildfowl 44, 75-100 (1993). sequences were ext racted and amplified in Washington DC , in a dedicated laboratory in 10. Giffin, J. G. 'Elepaio 35, 1- 3 (1993). 11. Worthy, T. H.J. Zoo/. 215, 619--028 (1988). a remote building, before studies were carried out in both DC and South Carolina on 12. Cooper, A. in Ancient DNA (eds Herrmann, B. & Hummel, extant taxa. There were 366 homologous positions for all taxa, and 36 variable sites S.) 149-165 (Springer, New York, 1993). within the ingroups. The subfossil sequences clearly group with t he Laysan duck, and 13. Kumar, S., Tamura, K. & Nei, M . MEGA : Version 1.01 are genetically distant from the koloa and mallards. (Pennsylvania St . Univ., 1993). 14. Swofford, D. PAUP: Version 3.1.1 (Illinois Nat. Hist. Survey, Champaign. 1993). 484 Nb.TURE · VOL 381 · 6 JUNE 1996