Giménez & Dick; Shore crab settlement and transport mechanisms 
161 
Bft 1 = 0.3 to 1.5 ms 2 = 1.6 to3.3 ms' 1 ,3 =3,4 to 5.4 m 
s’ 1 , 4 = 5.5 to 7.9 m s’ 1 , 5 = 8.0 to 10.7 m s’ 1 , 6 = 10.8 to 
13.8 m s’ 1 , 7 = 13.9 to 17.1 m s’ 1 , and 8 = 17.2 to 20.7 m 
s’ 1 : World Meteorological Organization 1970). The N-S 
and W-E components were regarded as positive when 
the wind was from the north or east, respectively. 
Daily residual surface currents (0 to 8 m), predicted 
from an operational circulation model developed by 
Federal Maritime and Hydrographic Agency for the 
(a) 
German Bight (Dick et al. 2001), were used to explore 
potential effects of current patterns on colonisation of 
megalopae. The model simulates tidal-, density- and 
wind-driven motion. Based on meteorological forecasts 
supplied by the DWD, the hydrodynamical model pre 
dicts currents in a nested grid system. The model had 
been previously validated in the North Sea with good 
agreement between data and predictions (Klein & Dick 
1999, Dick et al. 2001), and has been successfully used 
to predict and combat marine pollu 
tion (e.g. Dahlmann & Miiller- 
Navarra 1997). Current components 
were calculated from predicted sur 
face current fields in the German 
Bight with 1.8 km grid spacing. The 
data covering 50 km around Helgo 
land were coded in an 8-sector 
rosette. Current components are pos 
itive if currents are moving east 
wards or northwards. 
Data analysis. Time series analysis 
followed the protocol shown in 
Fig. 2a (after Chatfield 2004) using 
Statistica®. For abundance data, colo 
nisation rate of all collectors were 
pooled each day. Visual inspection of 
the time series showed trends and 
considerable variability, which in 
creased with increasing abundance. 
Therefore, colonisation data were 
log(x+l)-transformed, the trend re 
moved, and the data smoothed by a 3 
points moving-average. The function 
used for trend extraction depended 
on the year: for 2003 and 2004 data 
showed either an increasing or a 
decreasing trend, so a linear function 
was used; in 2005 data showed a 
humped pattern, and the trend was 
extracted using a log-normal function 
adjusted by non-linear regression 
(Fig. 2b). 
Spectral analysis of the de-trended 
time series of abundance was used to 
explore temporal patterns of varia 
tion. Relationships between abun 
dance, wind and predicted currents 
were explored using cross-correlation 
analysis after smoothing all series. Po 
tential relationships between abun- 
240 dance and tidal cycles were also 
explored with cross-correlation ana 
lysis, using the days since spring tide 
as variable. All cross-correlations 
were done by lagging the predictor 
Pooling colonisation 
X(f> = x, (í) + Xj (0 +x,(() 
Transformation 
V*{i) = log 1) 
I 
Trend subtraction 
Z{t) = Y(t)~Y(t) 
L 
Spectral analysis 
Wind or current components 
C(t) 
I 
Trend subtraction 
S (() = C (0 - C (f) 
Smoothing 
Cross- 
Smoothing 
¿(0 
correlation 
S(f) 
Spectral analysis 
(b) 
£Z 
o 
to 
(O 
c 
o 
o 
o 
<31 
o 
JO 
(0 
3 
•g 
m 
ai 
CT 
jo 
(0 
3 
■o 
(0 
01 
t_ 
to 
01 
■C 
5 
4 
3 
2 
1 
0 
2 
1 
0 
-1 
-2 
2 
O -1 
£ 
(O 
-2 
Ÿ (t) = 0.2598 it ■' e “ 27 - 5S72 (Log < - s.2430|< 
170 
180 
190 
200 
210 
220 
230 
Day 
Fig. 2. Procedure for time series analysis, (a) Flow diagram showing data trans 
formation for spectral analysis and cross-correlation; (b) example of data trans 
formation for colonisation by Carcinus maenas megalopae in 2005 (data from Fig. 6)