54- 
PACIFIC SCIENCE, Vol. XXI, January 1967 
coral was not available. Since the writer can 
distinguish between most live acroporid and 
pocilloporid corals on the basis of odor alone, 
he assumes here that chemical "odor,” or the 
absence of it, may cause these crabs to seek 
out their preferred host corals when they are 
present. 
Experiment 12 repeated Experiment 10 but 
arranged the corals in a definite upstream- 
downstream relationship within the four run- 
ning-seawater aquaria used. Each of the four 
aquaria had one acroporid coral with 4 Trapezia 
(a total of 16 Trapezia ), and one pocilloporid 
coral with 4 Tetralia (a total of 16 Tetralia). 
Corals were placed about 10 inches apart (be- 
tween the nearest opposing borders). In aquaria 
A and B the acroporid corals were placed up- 
stream to the pocilloporid corals. The converse 
was true with aquaria C and D. 
The results of Experiment 12 A and B, after 
24 hours were as follows: 6 Trapezia moved up- 
stream to their preferred pocilloporid corals, 
while 2 Trapezia remained on or under the 
acroporid coral ; 4 T etralia migrated downstream 
to their preferred acropid corals, while 4 were 
missing altogether. 
The results, of experiment 12 C and D, after 
24 hours were as follows: only 2 Trapezia 
migrated downstream to the pocilloporid corals, 
while 6 remained on or under the acroporid 
coral; 4 Tetralia migrated upstream to the pre- 
ferred acroporid coral, 2 remained with the 
pocilloporid coral, and 2 were lost. 
The combined results of experiment 12 show 
that, of the 16 crabs that did migrate, 10 
moved upstream to their preferred host while 6 
migrated downstream. Again, these results are 
not conclusive but suggest that chemical odors 
may enhance the location of the preferred coral 
host. 
In Experiment 13, 3 Tetralia specimens were 
placed in each of four large running-seawater 
aquaria which also contained some nylon mesh 
netting soaked with mucus from live acroporid 
coral. Only 4 of the 12 test animals located 
the mucus-gauze "bait” after 24 hours. These 
experiments were considered incomplete, how- 
ever, and will be continued in the future. 
To test more fully the reactions of Trapezia 
(Experiment 14, using T. /. ferruginea and T. 
f. areolata) in its host coral Pocillopora, crab 
specimens were starved for 3 days, then returned 
to live corals in small aquaria. When accus- 
tomed to the aquaria many crabs began what 
turned out to be feeding activities. The follow- 
ing is an account of the typical behavior dis- 
played : 
The crab climbed into the coral branches, 
then placed the dactyli of the walking legs 
(WL) 3 and 4 into polyp cups, depressing 
the polyps. Next, WL-1 were inserted into 
other polyp cups between the tentacles of the 
polyp, and "scratched” back and forth at a 
rate of about 4 strokes per second, for about 
4 seconds. The tips of WL-1 were then alter- 
nately cleaned by the mouthparts of the crab. 
During the cleaning operation, WL-2 were used 
to scratch new polyps, then were cleaned by 
the mouthparts. Mucoid material could be seen 
clinging to the tips of the walking legs during 
this procedure. 
Periodically material was transferred from 
WL-3 or 4 to WL-2 or 1, and then brought to 
the mouth. Occasionally the chelipeds were 
moved over the coral epidermis between the 
polyps. The fingers subsequently were cleaned 
in the mouthparts of the crab. These activities 
were repeated over and over again, as the crab 
slowly moved up through the coral branches. 
Upon examining the dactyli of Trapezia f. 
ferruginea , it was noted that a special brush 
and comb is present on the terminal segments of 
each leg (referred to henceforth as the food 
brush and food comb). The food brushes are 
situated at the distal end of each dactylus (Fig. 
1 E) and consist of several short, stout, blunt- 
ended spines for agitating the coral polyp, and 
a dense tuft of bristles for collecting mucus, 
bacteria, and other debris. The bristle tuft is 
fully developed in walking leg 1 but is pro- 
gressively less well represented posteriorly (Fig. 
1 A-D). Borradaile (1903:240) figures the 
terminal spines of Trapezia f. ferruginea and 
suggests that "the remarkable ending of its legs 
is in some way connected ...” with its life 
in the coral. The terminus of each dactylus 
protrudes ventrally, thus forming a concavity 
on the ventral surface of the immediate proxi- 
mal part of the dactylus. The food combs, shown 
in the posterior view of the left dactyli (Fig. 
1 A-D), consists of from 3 to 6 rows of 
