300 
PACIFIC SCIENCE, Vol. XVIII, July 1964 
rough, but, for the purposes of the present paper, 
adequate method of obtaining comparable data 
is to extrapolate from the original curve plotted 
from the data in Table 1. This has been done 
in Figure 2 C, with the original curve being 
evaluated at 7-day intervals. 
Andrewartha and Birch (1954) have dis- 
cussed the innate tendency of animals to dis- 
perse, but apparently they attached little general 
significance to dispersal in response to crowding. 
Such a mechanism has been shown for the 
great tit by Kluijver (1951) and it is known 
that in some crab species dispersal is directly 
related to population density (Bovbjerg, I960). 
It is obvious from Figures 2C and 2D that there 
are strong dispersal movements into the un- 
occupied area within a relatively brief period. 
It is not obvious, however, whether this dispersal 
is in response to the density of the individuals 
outside the sampling area. To examine this 
problem, two alternates may be erected: 
Alternate 1: Dispersion remaining constant 
and independent of density. Assuming that the 
dispersion of pagurid populations is constant 
and independent of their density, all crabs would 
spend a great deal of their time wandering 
about. If the individuals found in a grid, such 
as that erected at Eniwetok, were removed at 
constant intervals, the total size of the pagurid 
populations would decrease at a constant rate, as 
is hypothetically depicted in Figure 2 A (curve 
a). This pattern assumes both density inde- 
pendence and a status quo, with no death or 
recruitment to the population. Viewed from the 
number of individuals collected in the grid, 
however, a different pattern would result from 
the same assumptions. Since the total size of 
the pagurid populations is very large, compared 
to those collected in the grid, the number of 
individuals collected at each interval would not 
vary to any great extent. The result (Fig. 2 B, 
curve a) would be a curve which was parallel 
to the abscissa or which decreased linearly to a 
very slight extent. Continued for an extremely 
long time, the curve would approach that ob- 
tained by viewing the entire area, since every 
crab would eventually wander across the grid. 
When crabs ceased to enter the grid, the ex- 
planation required by density-independent dis- 
persal is that all crabs have been collected. 
Alternate 2: Dispersion as a function of den- 
sity. Assuming that the tendency to disperse 
increases with increasing population size and 
conversely decreases with decreasing population 
size, a different pattern would result. The num- 
ber of pagurids in the entire area (Fig. 2 A, 
curve b) would initially drop at a rate greater 
than the final rate. The curve then indicates a 
variable rather than a constant-rate function. 
At the threshold of density-dependent dispersal 
effects, that is, at the point above which crabs 
disperse and below which they do not, the 
number of crabs obtained in the sampling grid 
(Fig. 2 B, curve b) would be 0. However, in 
contrast to density-independent dispersal, there 
would still be pagurids in the area ( as depicted 
in Fig. 2 A, curve b). 
A final evaluation of these two alternates is 
not possible in the present study. For example, 
surf action may be responsible for most of the 
influx at Horseshoe Cove, and no marking and 
recapture experiments were made at either site. 
However, there are indications that Alternate 
2 more adequately explains the data. If the im- 
migration rates of the three Eniwetok crabs are 
examined, either lumped or separately (Fig. 2C), 
it is apparent that the pattern more nearly ap- 
proximates that required for density-dependent 
dispersal than it does for density-independent 
dispersal (Fig. 2 B). Variations in the slopes of 
the collecting patterns may possibly reflect dif- 
ferences in the initial density levels relative to 
the threshold of density effects. 
Repopulation of two Horseshoe Cove pools 
( Fig. 2D ) also agrees more closely with Alter- 
nate 2 than with Alternate 1. 
SHELL UTILIZATION 
Shells housing individual pagurids were iden- 
tified whenever possible. Each pagurid species 
was found to utilize the shells of different gastro- 
pods with different frequencies in both the 
Eniwetok and Californian studies (Tables 2 
and 3). 
To test for homogeneity of shell utilization 
in various samples of the same species and to 
inspect the collections for interspecific differ- 
ences in shell utilization, a trellis diagram was 
constructed (Fig. 3). For a discussion of the 
mechanics of this type of test, the reader is 
referred to Wieser ( I960) and Sanders ( I960) . 
