8.6.2 Empirical evidence for regime shifts in marine ecosystems
At the population level, Collieet al.(2004) applied
Scheffer and Carpenter’s (2003) criteria to test for
regime shifts in Georges Bank haddock, based on
age-structured abundance time series for the period
1931–2000. Although some of these criteria are dif-
ficult or impossible to evaluate in open marine eco-
systems (e.g. whether the system goes to different
states after a perturbation or under different start-
ing conditions), others can be addressed with avail-
able time-series data (Fig. 8.8b–e). First (Fig. 8.8b),
they demonstrated a discrete step in the average
stock biomass, which dropped abruptly after a
spike in catch due to influx of foreign fishing fleets
in the early 1960s, and remained low for most of the
remaining century. Second (Fig. 8.8c), the distribu-
tion of biomass values was bimodal. Third (Fig.
8.8d), catch showed a different functional relation-
ship to fishing mortality before and after the shift in
the 1960s. These three observations support the
existence of two distinct regimes during the time
series. Moreover, simulations of a population
model fit to empirical data showed that catch fol-
lowed different trajectories when fishing mortality
was increased versus decreased, i.e. the system ex-
hibited hysteresis (Fig. 8.8e), a key piece of evidence
for a discontinuous regime shift between alternate
semi-stable states (Collieet al. 2004).
An ecosystem-wide regime shift, evidently
forced in part by overfishing, has also been docu-
mented in the Black Sea (Daskalovet al. 2007). Here
strong fishing pressure on top predators caused
their decline and eventual collapse in the 1970s,
which was accompanied by cascading changes in
lower trophic levels, leading to phytoplankton
blooms and nutrient depletion; a subsequent
change in the focus of fisheries to smaller plankti-
vorous fishes such as sprat and anchovy (‘fishing
down the food web’; Paulyet al. 1998) then led to a
subsequent collapse of these planktivores and
corresponding increases in the jellyfish that com-
pete with them (Daskalov 2002). Plotting the time
trajectories of the various trophic groups shows
that, for several interactions, the relationships be-
tween consumer and prey abundances differed in
early compared with later years, suggestive of the
hysteresis characteristic of a discontinuous regime
shift (Fig. 8.8f–h). Models confirm that overexploi-
tation can trigger such shifts between alternate
states (Daskalov 2002; Collieet al. 2004).
Are such regime shifts common in marine eco-
systems and, if so, how do they relate to forcing
mechanisms? Fenget al. (2006) explored this ques-
tion using dynamic simulation of mass-balance
models (Ecopath with Ecosim; Christensen and
Walters 2004) applied to 24 marine ecosystems.
The models imposed a simulated perturbation of
10 years of intensified fishing, then relaxed fishing
to the initial level and followed the system’s long-
term (70 years) trajectory, asking whether each sys-
tem returned to its original state or equilibrated to
an ‘alternative attractor’. Six scenarios considered
fishing on top predators compared with intermedi-
ate levels (wasp-waist system) under bottom-up,
mixed or top-down control. The simulations
showed that, under top-down or mixed control,
11–28% of the ecosystems showed alternate attrac-
tors, i.e. shifted into a new regime that persisted
after fishing pressure was relaxed, whereas none of
the ecosystems showed alternate attractors under
bottom-up control (Fenget al. 2006). These model
results, together with a few well-documented em-
pirical examples such as the Black Sea, suggest that
intense fishing pressure can produce shifts to new
ecosystem states that are difficult to reverse, sup-
porting suspicions that such regime shifts may be at
least partly responsible for the failure of many
heavily fished species to rebound even decades
after fishing moratoria were enacted (Hutchings
and Reynolds 2004). If such regime shifts are
indeed common responses to top-down perturba-
tions, they have serious implications for mana-
gement of natural ecosystems.8.6.2.1 Mechanisms
Several mechanisms potentially can produce re-
gime shifts between alternate attractors or semi-
stable states (Collieet al. 2004; Folkeet al. 2004;
Knowlton 2004). At the population level, the most
general mechanism involves the Allee effect, i.e.
depensation or positive density dependence at
low population sizes (Knowlton 1992), although
this process requires some additional factor to112 APPLICATIONS