florisil, and keeping the extract frozen until just before it was 

 injected into the HPLC. In 3 crudely quantified HPLC assays that 

 we performed immediately after collecting tips from H. 

 incrassata , we found halimedatrial concentrations ranging from 

 2-6% of blotted wet mass. These concentrations are higher than 

 our rigorously quantified measurements but are similar to 

 immediately analyzed extracts of Halimeda from the Pacific (Paul 

 and Van Alstyne, in press). Thus, our concentration data may be 

 low but the observed patterns of relative concentration would not 

 have been affected since all samples were treated similarly. 



To evaluate the susceptibility of young versus old Halimeda 

 segments to herbivores, we paired branches of Halimeda incrassata 

 having newly formed apical segments (approximately 20 hrs old) 

 with branches having apical segments that were more than 48 hrs 

 old (the shortest time in which we ever saw a new segment come to 

 resemble an old one) and transplanted these onto the reef slope 

 at a depth of 15 m. On our "young" branches, the top 2-3 

 segments appeared to have been produced recently as evidenced by 

 their bright green color and soft texture, which appear to be a 

 consequence of their incomplete calcification. All of the 

 segments on our "old" branches appeared to be fully calcified. 

 Each branch was 5 segments in length. The basal segment of each 

 branch was wedged between the strands of a length of 3-strand 

 rope and ropes were attached to pieces of coral on the reef. A 

 distance of 2-3 m separated each rope. Thirty-seven ropes were 

 placed on the reef at 1200 hrs and retrieved at 1000 hrs the next 

 day. Grazing was measured as the number of segments consumed. 

 Herbivores removed all of both plants on 2 ropes and none of 

 either plant on 17 ropes; these replicates were excluded from the 

 paired-sample analysis since they provided no information on the 

 relative susceptibility of new versus old branches. 



A similar test using branches of Halimeda simulans was 

 performed in the 2-3 m deep bay near the Hydrolab shore base. 

 Thirty-eight ropes with paired branches were set out at 1800 hrs 

 and retrieved at 0930 hr the next day. At this site grazing was 

 slight and was due primarily to very small scarids that did not 

 remove entire segments. Here we counted the number of new versus 

 old apical segments showing bite marks. To identify the fishes 

 responsible for grazing in this habitat, we took time lapse 

 movies (1 frame/min) of a dense patch of H_^ simulans from 1500- 

 1000 hrs using an underwater 8 mm camera and electronic strobe 

 mounted on a tripod. 



We tested the effects of halimedatrial on herbivorous fish 

 feeding by comparing the consumption of palatable plants coated 

 with either (1) halimedatrial dissolved in diethyl ether or (2) 

 diethyl ether alone (i.e., a control). Two palatable seaweeds 

 were used, the seagrass Thalassia testudinum , which is preferred 

 by parrotf ishes, and the red alga Acanthophora spicif era , which 

 is preferred by surgeonf ishes (Lewis, 1985). Using these two 



373 



