Geoscience Reference
In-Depth Information
The default parameter values used for these runs are summarised
in
Table 1
.
Our inspiration for the final scenario was drawn from the phytoplankton
blooms that occur in frontal regions, such as the Celtic Sea shelf edge,
producing dynamic areas of high productivity (
Batten et al., 1999
). The
spatial scales of processes in this scenario exceed those considered in the
other simulations, so we increased the extent of our grid to represent an area
of 100
10 km. The simulated phyto-
plankton bloom was assumed to begin near one spatial boundary, growing in
peak biomass density at a rate of 0.084 g m
3
yr
1
over a period of 2 months
(
Joint et al., 1986
) and dying out over the rest of the year. Because the
phytoplankton were distributed across the sizes 0.8
500 km with a grid resolution of 10
g to 0.9 mg, this was
achieved by increasing the intercept from 0.006 to 0.02 g m
3
consistent with
values reported in field studies and used in biophysical models (
Blanchard
et al., 2009; Reul et al., 2005
). The slope of the phytoplankton size spectrum
was assumed to remain constant. The bloom begins as a Gaussian distribu-
tion in space with 95% of the phytoplankton density spread across an area
with a diameter of ca. 20 km, approximating a mesoscale phytoplankton
bloom. The current speed was taken as 1200 km yr
1
, and intermediate value
for speeds in surface waters in the Celtic Sea (
Pingree et al., 1981
).
Outputs from all runs were presented as 3D surface plots showing the
spatial distributions of defined size classes through time (or the average
distribution over a given period). Outputs were also used to generate size
spectra for different locations and generalised 'growth' trajectories in a 4D
mass-space-time phase space. These trajectories tracked the movement and
growth of individuals of 5 mg initial mass from different starting locations.
The individuals were tracked until they stopped growing or attained upper
mass boundary for the model (1200 kg).
m
G. Data
We compared the modelled spatial size spectra with empirical patterns from
the Celtic Sea fish community. Abundance data by size classes and species
were collected from the annual CEFAS Celtic Sea groundfish surveys. The
data were standardised to account for differences in trawl durations and
therefore are indices of relative abundance (number of individuals per hour).
Only locations consistently sampled through the time series over the years
1987-2001 with the Portuguese high-headline trawl were used (
Blanchard
et al., 2005a
). The size spectra were calculated as the logarithm of relative
abundance of all individuals summed within log body-mass classes. To
examine spatial patterns, we time averaged over the 1987-2001 period and
examined size spectra by sampling location.
