been previously attempted. This visit was part of the Third 

 Joint US-USSR Bering & Chukehi Seas Expedition, a 

 4-month voyage that also ineluded oceanographic studies in 

 the Pacific Ocean and South China Sea. 



In 1990, one of the authors (AKK) visited Caroline twice 

 with the International Council for Bird Preservation (ICBP) 

 1990 Line and Phoenix Islands Expedition. These visits were 

 primarily to discuss conservation matters with the Falconers; 

 introduce Caroline to Dr. and Mrs. M. Garnett, representatives 

 from ICBP; confirm the illegal taking of fish, turtles, and 

 coconut crabs by Mr. Urima's men; find the Tuamotuan 

 maraes; survey North Arundel Islet; collect invertebrates; and 

 fill in missing details from the 1988 expedition. 



Methods 



From 22 to 29 September 1988, Drs. A. K. Kepler, C. B. 

 Kepler, D. H. Ellis (USA), and Mr. Katino Teeb'aki (Republic 

 of Kiribati) surveyed all 39 motus at Caroline Atoll (Fig. 2), 

 gathering detailed information on plants, seabirds, land birds, 

 mammals, reptiles, coconut crabs, and human disturbance. 

 Some incidental data have been added from the two visits in 

 1990 (10-13 March, 18-28 May) by Dr. A.K. Kepler, Captain 

 G. Wragg, A. Garnett, M. Linsley. J. Phillips (March), and 

 Dr. M. Garnett (May). 



Prior to the first expedition, a series of transects and known 

 botanical information were mapped to ensure that 5% of each 

 motu was sampled and to maximize our chances of encountering 

 all known plant species. Transects on the larger motus (Nake, 

 Long, South ) were spaced approximately 400 m apart and, with 

 one exception, were perpendicular to the long axes of each islet 

 (Tr. 3 on Nake extended first from east to west, then ran south 

 parallel to the west coast). If the motus were greater than 

 400 m long, we used two transects. Transects on the smaller 

 motus passed through their widest points (Fig. 8). Their 

 lengths ranged from 77 m (Azure) to 2,000 m (Tr. 3, Nake). 



Considerable modifications were required when we realized 

 that the Clapp& Sibley ( 197 la) map (Fig. 7) was incorrect. We 

 redrew the transects on Arundel's 1883 map (Fig. 4), secured 

 just prior to the expedition. On South Island, due to impenetrable 

 draperies of Ipomoea vines, Tr. 3 was omitted, Tr. 5 ran only 

 from the lagoon south to the Ipomoea curtain (75 m), and 

 Trs. 4 and 6 ran north and south until we reached an impasse 

 (PI. 7). 



Compass headings were determined by the configuration 

 of each island. Beginning at high water mark, all distances 

 (islet dimensions, widths of reef flats and substrates, and plant 

 communities) were measured using hip chains with 

 biodegradable cotton thread. These parameters were later 

 checked against aerial photographs in stereoscopic pairs taken 

 in 1 985 by the Royal New Zealand Air Force ( RNZAF), which 

 provided 3-dimensional overviews of every islet. Vegetation 

 maps for each islet were constructed by drawing the outlines 

 and plant communities (based on the aerial photos) on graph 

 paper, enlarging, then counting dots. 



Data were collected in a 30-m (15 m to each side of the 

 transect) swath along each transect and recorded on field 

 forms. Within each plant community we took photographs. 



assessed the relative abundance of each plant species (rare, 

 uncommon, common, very common, abundant, and locally 

 common; see Vegetation section ) and recorded plant community 

 width, plants collected, and substrate type. We also estimated 

 the maximum height of the dominant vegetation and percentage 

 of ground area covered by each species. 



In addition to the linear transects, an additional 19,300 m 

 of perimeter surveys were conducted on 21 islets (Fig. 9). The 

 combined distance for linear and perimeter transects was 

 32.6 km. Seven tiny islets (Noddy Rock, Skull, Atibu, Bosun 

 Bird, Coral, Reef-flat, Fishball) were surveyed completely. 



In 1988. we camped on the atoll for seven nights, 

 establishing base camps ( Fig. 8; Pis. 6,8) on the northwest point 

 of South (22-24 September) and southwest Long 

 (25-28 September). We relocated camps using a Zodiac with 

 outboard motor and an inflatable Sevylor canoe (PI. 9). All 

 transects were surveyed during daylight hours, beginning at 

 dawn. Walking the interislet channels was relatively easy at 

 low tide, but became hazardous at incoming or high tide, as the 

 numerous black-tipped reef sharks, Carcharhinusmelanopterus 

 (PI. 10), regularly charged at our legs and had to be beaten off 

 with sticks and coral chunks. 



During the 1990 visits, we stayed on Motu Ana- Ana with 

 the Falconers. Work was not intensive, as in 1988. We walked 

 or motored an inflatable Lancer, visiting 20 motus: Ana-Ana, 

 Kimoa, Pisonia, Eitei, South, North Arundel, Noddy Rock, 

 Brothers, North Brothers, Skull, Pig, North Pig, Bo'sun Bird, 

 Long, Nake, Mouakena, Shark, Scarlet Crab, Bird, and Fishball. 

 Insects were preserved in ethyl alcohol. We used a "Magellan" 

 NAV 1000 to obtain accurate geographical coordinates of 

 Caroline. 



Naming Caroline 's Motus 



Previous literature has provided vague or incomplete data 

 on Caroline's constituent motus (Bennett, 1840; Holden & 

 Qualtrough, 1884;Markham, 1904; Stevens &Barwick, 1930; 

 Bryan, 1942; Clapp & Sibley. 1971a; Garnett, 1983). This 

 confusion resulted because most previous visits had been brief. 

 The only charts available were a quite accurate survey by 

 Arundel, a guano merchant who mapped the atoll in 1883 

 ( Fig. 4; Admiralty Chart, No. 979, 1 965 ), and a map. greatly in 

 error, drafted by an international Solar Eclipse Party, also in 

 1883 (Fig. 5). There are no hydrological navigation charts. 

 Unfortunately, the eclipse party map has been used in all 

 subsequent scientific, historical, and sociological publications 

 (Bryan, 1942 [Fig. 6]; Maude. 1968; Clapp & Sibley, 1971a 

 [Fig. 7]; Garnett, 1983). It shows only 25 of the 39 islets and 

 many of the shapes are distorted. The 38 islets on Arundel's 

 map are similar to those in the RNZAF (1986) aerial 

 photographs. Only a few islets appear to have changed in 

 minor ways since 1883: major discrepancies in Arundel's map. 

 we believe, are due to difficulties involved in the accurate 

 rendition of small land areas (i.e., the South Nake Islets). 

 Maude (ca. 1938) counted 36 islets but never published his 

 information. 



To aid our survey we named 32 islets and 4 islet groups 

 (Fig. 2). Our names, whether in English or Gilbertese, reflect 

 appropriate aspects of islet biology. Etymology is provided in 



