454 
Fishery Bulletin 106(4) 
quickly after establishment of a closed-area because the 
life span of individual colonies typically does not exceed 
one year (Hughes, 1989). The subsequent decline in 
the percent cover of this taxon may be linked to either 
increased interspecific competition or greater predation 
pressure at site 17. Organisms with life spans of five 
years or greater (i.e., sponges, P. magellanicus, Pagu- 
rus spp., Asterias spp., and <S. droebachiensis) either 
increased throughout the duration of this study at site 
17 or maintained a heightened level of abundance for 
several years before eventually experiencing a decline 
in numbers. Because differences between areas inside 
and outside CA-II, in terms of both sponge and bushy 
bryozoan cover, became apparent in 1997 (as shown by 
the error bars in Fig. 2), this finding may indicate that 
there is an approximately two-year lag in the initial 
response of colonial epifauna residing in gravel habitat 
to the fishing closure. This lag implies that even in- 
frequent trawling and dredging can lead to prolonged 
changes in colonial epifauna composition. 
Although no other research on Georges Bank has 
examined long-term patterns of recovery from distur- 
bance among epifauna, our results are similar to those 
of two studies conducted at Cashes Ledge (located 
130 km east of Gloucester, MA) on the colonization 
of artificial and disturbed natural substrate by colo- 
nial organisms over a period of less than two years 
(Sebens et al., 1988; Witman, 1998). In an area with 
rocky substrate at 30 m depth, the seafloor was first 
colonized by erect bryozoans ( Crisia eburnea and Id- 
midronea atlantica) and the polychaete Spirorbis spp., 
while later colonizers included the ascidians Aplidium 
pallidum and Ascidia callosa (Witman, 1998). At the 
end of the 15-month duration of the Witman study, 
bryozoans covered approximately 35% of the disturbed 
patches of the seafloor, whereas no other species had 
yet colonized more than 6% of this area. It was pre- 
dicted that it would require seven years or more for 
this epifaunal community to return to its natural state 
where sponges, sea anemones, and ascidians were the 
dominant fauna (Witman, 1998). In a similar study 
where four depth strata were examined, bushy and 
encrusting bryozoans were again the first organisms 
to settle on barren substrate; smaller patches of F. 
implexa and Spirorbis spp. were also early colonizers; 
colonization by Aplidium sp. and crustose coralline 
algae occurred later (Sebens et al., 1988). At the 50-m 
site on Cashes Ledge, which is the most comparable to 
our study area in CA-II, 1. atlantica covered over 50% 
of the substrate within one year, and the cover of bryo- 
zoans approached 100% by the second year. Although 
bryozoan cover at our study sites never reached the 
extremely high percentages observed by Sebens et al. 
(1988), our results mirror theirs and those of Witman 
(1998) in that bryozoans were the predominant space 
holder in areas subjected to recent disturbance, and 
all other species of colonial epifauna never exceeded 
a mean cover of 5%. The recovery of the benthic com- 
munity at Cashes Ledge may differ somewhat from 
Georges Bank, because 1) certain epifaunal species 
may prefer the rocky substrate at Cashes Ledge over 
the gravel substrate of Georges Bank (or vice versa); 2) 
different currents in these areas may affect the settle- 
ment rate of invertebrate larvae, and 3) the spatial 
scale of disturbance at Cashes Ledge (i.e., 25 cm 2 ) was 
much smaller than at Georges Bank. 
Some evidence indicates that ecological succession is 
governing the recovery of epifauna in CA-II, but several 
generations of data on key benthic species would be 
needed to conclusively rule out other possible mecha- 
nisms controlling recovery. There are two models that 
describe how marine ecosystems may recover from past 
bottom-fishing disturbance (Auster and Langton, 1999): 
1) Ecological succession may occur where the seafloor 
is colonized by benthic species in a predictable pattern, 
namely opportunistic species arrive first only to be later 
out-competed by climax species; 2) The lottery hypoth- 
esis, first proposed by Sale (1978), indicates that the 
benthic community will be dominated by species whose 
larvae are the first to colonize the seafloor immediately 
after a disturbance. This scenario implies that recovery 
will be greatly influenced by unpredictable, stochastic 
events. Because many species in CA-II have life spans 
of ten years or more (e.g., the Northern seastar Asterias 
vulgaris, B. undatum, P. magellanicus, S. droebachien- 
sis-, Hermsen et al., 2003), it is not yet possible to use 
time-series data to quantitatively differentiate between 
these two hypotheses. However, there are indications 
that recovery from physical disturbances is not char- 
acterized by as much stochasticity as implied by the 
lottery hypothesis. This lack of stochasticity is implied 
by the fact that bryozoans are early colonizers at both 
CA-II and Cashes Ledges. Another tenet of the lottery 
hypothesis is that early colonizers are able to consis- 
tently out-compete late colonizers by denying them ac- 
cess to vital habitats (Munday, 2004). This corollary 
of the lottery hypothesis is not met on Georges Bank, 
because encrusting bryozoans can be overgrown by the 
colonial tunicate Didemnum sp. and sponges (Valentine 
et al., 2007). 
After examining six years of data collected since 
the establishment of CA-II, we do not know whether 
recovery of colonial epifauna is complete at site 17. 
Although large increases in the cover of colonial epi- 
fauna were not seen during 1999 and 2000, the abun- 
dance of several noncolonial species continued to grow. 
This continued growth of noncolonial species could in 
turn affect colonial epifauna composition as recover- 
ing noncolonial organisms may increasingly compete 
with colonial taxa for resources or prey more heavily 
on them. The recovery of some colonial species may 
still be constrained by the fact that their larvae only 
disperse short distances (<50 m) from parent colonies 
(Hughes, 1989). The final species composition of site 
17 is likely to differ from that of undisturbed site 20, 
because the faster current speeds at the shallow site 17 
can transport sand across the gravel pavement, inhibit- 
ing the growth of arborescent epifauna and F. implexa. 
In fact, the scouring caused by sand transported to 
site 17 during a winter storm that coincided with our 
