696 



Fishery Bulletin 90(4). 1992 



X X X X Xj^ j^ 



X X X x'^x x'*x''x''x''x X X X X X 



X X X X X X „_^_ XXX XXX 



xxxxxxxxx REEF xxxxxxxxxxx 

 xxxxxxxxxx xxxxxxxxxxx 



X X XXXxX XXXXXXX x X X X X XXX 



15 m 



15 m 



NC 



in^'^'-r 401 



NC 



15 m 



NC 



15 m 



NC 



40 m 



iNC 



SHORELINE 



Figure 4 



Schematic of casita layout at the outer-bay site for the August 

 1989 experiment. S = small casita, M = medium casita, L 

 = large casita, and NC = no-casita station. 



assess predation in an environment uninfluenced by the 

 casitas. This assumption held true for the diurnal pred- 

 ator guild during August 1989 (see Results). Only juve- 

 nile lobsters approximating the small size-class {x 

 53.2mmCL, range 45.2-59.0mmCL, N 72, SD 4.1) 

 were tethered for 7 days. We chose only small lobsters 

 in the second experiment because logistical considera- 

 tions limited us to one size-class, and we wanted to 

 verify that survival of small lobsters was enhanced 

 when residing beneath casitas (See Results for first ex- 

 periment below). Each casita and no-casita (i.e., metal 

 frame) station contained 6 tethered lobsters for a total 

 of 72 tethered lobsters (6 lobsters x 4 distances (0, 15, 

 30, and 70 m) x 3 replicate stations). 



Table 1 



(a) Three-way ANOVA table (model I) describing the effects 

 of site (inner-bay sand-seagrass flat, and outer-bay seagrass 

 bed adjacent to coral reefs), lobster Pamdirus ar^us size (small 

 46-55 mm CL; large 56-65 mm CL) and shelter availability 

 (casita vs. no-casita station 60 m away) on proportional mor- 

 tality rates (arc-sine square-root transformed) of tethered 

 lobsters during January 1989. •P<0.05, **P<0.01. ns 

 P>0.05. 



(b) Ryan's Q tests of mean proportional mortality rates (arc- 

 sine square-root transformed) of tethered lobsters for the in- 

 teraction effect of lobster size x shelter availability. Treat- 

 ment levels not significantly different at the 0.05 level share 

 an underline. Treatment levels are arranged in increasing 

 order of proportional mortality. 



Interaction 



Shelter availability 



Casita 

 No Casita 



Lobster size 



Small 

 Large 



Lobster size 



No Casita 



Casita 



Lobsters were checked and predation losses scored 

 every 1-2 days during experiments. Fewer than 4% 

 of tethered lobsters escaped, and these were not used 

 in subsequent statistical analyses. Lobsters that were 

 eaten or missing were not replaced. Cumulative losses 

 were converted to proportional mortality/day/casita (or 

 station). Proportions were analyzed as a function of 

 shelter availability (casita vs. no casita), distance from 

 the casita (0, 15, 30, and 70 m), lobster size (small vs. 

 large), and site (inner-bay vs. outer-bay) with two- and 

 three-way, fixed-factor analyses of variance (ANOVA) 

 models (after procedures in Underwood 1981). Propor- 

 tional mortality was arc-sine square-root transformed 

 to meet assumptions of normality and homogeneity of 

 variance (Underwood 1981). In all cases, the variances 

 were homogeneous as determined by Cochran's C-test. 

 Differences among means were revealed by use of 



