2.2 1980s to present: Integrated Trend Analysis

Temporal development 2016
Typography
  • Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times

PCA – Abiotic variables

A Principal Component Analysis (PCA) for all abiotic variables, as listed in Table (Appendix 1), has been performed (Figure 2.2.1). Clearly, the analysis reveals that the two main drivers of variability of the Barents Sea climate system is the overall temperature (PC1; explaining 47% of the variability) and the through-flow of Atlantic Water (PC2; explaining 20% of the variability). The dynamic variability, i.e. changes in volume transports, are associated with changes in the North Atlantic Oscillation (NAO).

The maximum sea ice cover (April) is strongly correlated with the temperature in the Kola Section, while the analysis shows that the changes in the maximum sea ice cover (September) is related to both the temperature and the volume transports in the Barents Sea. Notable years are easily distinguishable; both 2016 and 2015, as well as 2006, 2007, and 2012 were all anomalously warm and hence, are placed near the positive end of the x-axis (PC1). The years of the so-called "Great Salinity Anomaly" (1978, 1979, and 1980) are at the negative extreme of the x-axis. The areas of Atlantic Water and Arctic Water are positively and negatively correlated, respectively, with PC1.

Interestingly, 1979 and 2010 both belong to the negative extreme in the dynamical dimension, i.e. PC2, but they are far apart on the temperature scale (x-axis) as they belong to two different climatic regimes of the Barents Sea, namely the cold 1970s and the warm 2000s, respectively. 1989 stands out as the most extreme year in the dynamical dimension, marking the start of the current warm period.

The third axis (PC3; not shown) is related to the variability of the Atlantic Water inflow through the Barents Sea and is negatively correlated with the minimum sea-ice cover in September. We interpret this as the dynamic part of the summer sea-ice minimum, which is often close to zero and therefore less dependent on the temperature and, according to the analysis, also dependent on the dynamic state of the Barents Sea. The summer sea-ice minimum is also negatively correlated with the exchanges through the openings in the northern Barents Sea.

Figure 2.2.1 PCA of abiotic variables for the period 1979–2016Figure 2.2.1 PCA of abiotic variables for the period 1979–2016

Figure 2.2.2 clearly shows the «borealization» of the Barents Sea during the period 1979–2016. Temperature related variables, such as temperature itself, the area occupied by Atlantic Water (T >3°C), and the ice area at minimum and maximum sea-ice extent all show a clear trend towards warmer conditions. The variables connected to the dynamics, i.e. the flow of Atlantic Water (e.g. BSO – the inflow of Atlantic Water to the Barents Sea) does not show any clear trend, but rather some fluctuations partly in correspondence with the atmospheric forcing represented by the NAO. These results corroborate the findings that, while both the temperature and volume transport of the Atlantic Water into the Barents Sea contribute to the oceanic heat transport into the Barents Sea, they (temperature and volume transport) vary on different time-scales. While the volume transport fluctuates from year to year and even shorter time-scales, the temperature shows a clear signal of multidecadal variability, i.e. climate variability of addition to an underlying trend induced by climate change.

Figure 2.2.2. Time-series of annual means of abiotic variables in the Barents Sea sorted by trend. Blue means value below the 1979–2016 average and red means above the 1979–2016 average.Figure 2.2.2. Time-series of annual means of abiotic variables in the Barents Sea sorted by trend. Blue means value below the 1979–2016 average and red means above the 1979–2016 average.

PCA – Zooplankton and 0-group fish

A PCA analysis was done on a set of variables including zooplankton biomass in three size fractions and sum total for the Barents Sea (Figure 2.2.3), 3 time-series of krill, and abundance of 0-group fish of 8 species (capelin, cod, haddock, herring, polar cod, Greenland halibut, long-rough dab, redfish, and saithe). These are all organisms that drift (or have drifted) with the currents into and inside the Barents Sea.

Figure 2.2.3. Results from PCA on 16 biotic variables: zooplankton biomass, abundance indices of krill, amount of jellyfish, and abundance of 0-group fish. Biplot of variables and years plotted on PC-1 and PC-2.Figure 2.2.3. Results from PCA on 16 biotic variables: zooplankton biomass, abundance indices of krill, amount of jellyfish, and abundance of 0-group fish. Biplot of variables and years plotted on PC-1 and PC-2.

Three of the zooplankton biomass variables fell in the lower right-hand corner (total, small and medium size fractions), while the three krill variables pointed downwards associated with negative values along PC-2. Capelin, cod and redfish 0-group pointed in the same direction, whereas 0-group of herring and saithe pointed in the opposite direction. 0-group of Greenland halibut and polar cod were aligned with positive direction of PC-1, whereas 0-group of long rough dab fell in the ‘southeast’ sector aligned with the medium size fraction of zooplankton biomass.

Figure 2.2.4. Same as Figure 1.2.3 but with environmental (abiotic) variables projected as passive variables onto the biplot.Figure 2.2.4. Same as Figure 1.2.3 but with environmental (abiotic) variables projected as passive variables onto the biplot.

When the set of abiotic variables were projected as passive explanatory variables onto the PCA biplot for the zooplankton, krill and 0-group fish, they aligned along an axis rotated about 30° to the right relative to the vertical PC-2 axis. The distinction between the warming and dynamic axes of the PCA for the abiotic variables (Figure 2.2.1) was less pronounced in this representation where the space was to some extent collapsed into a plane. The warming trend, revealed by the variables for areas of Atlantic and mixed waters and temperatures at the BSO and Kola sections, coincided with the variables for krill (total and southern Barents Sea) and 0-groups of capelin, cod, and redfish. In the opposite direction, associated with the cold side of the trend revealed by area of Arctic water and winter ice extent, were 0-group herring and saithe. These patterns reflected generally increasing trends with good recruitment of cod and capelin in the recent warm years, whereas recruitment of herring and saithe have been generally poor.

The projection of the years shows the low zooplankton biomass years in the early 1990s (1990–1992) in the upper left side, while the peak zooplankton biomass year 1994 is an extreme on the lower right (Figures 2.2.3 and 2.2.4). The recent warm years after 2006 are all located in the lower left part of the plot (except 2015 which is slightly on the right). 2004 is another extreme year on the opposite side in the upper-right corner.

PCA – Fish variables

A PCA analysis was run with 23 variables reflecting stock size, growth and maturation of cod (7 variables), haddock (6 variables), capelin (5 variables), polar cod (2 variables), and herring, long-rough dab and shrimp (1 variable each). A biplot with variables and years shown on PC-1 and PC-2 is given as Figure 2.2.5. In this case the environmental variables retained more of their configuration when projected onto the biplot, with the ‘warming trend’ variables oriented along PC-1 and the ‘flux’ variables oriented along PC-2 (Figure 2.2.6).

The fish variables formed several ‘bundles’ or clusters on the biplot. The capelin stock size grouped with weight-at-age of cod (3 and 5 years old) in upper right hand side of the plot where also some years with peak capelin stock (1990–1992) were projected. Another group of variables clustered around 4 o’clock in the diagram, consisting of the stock sizes of cod and haddock (3+, and haddock SSB). In the opposite direction were cod maturing at age 7 and cod weight-at-age as 8 years old, as well as shrimp mortality. The stock of juvenile herring (1–2 years old) was also oriented in this opposite direction. The recent years after 2007 grouped in the lower left portion of the diagram along with the variables for cod and haddock stocks.

Another group of variables that clustered in the lower left part of the biplot were related to growth of capelin as well as fishing mortality of cod and haddock. Years which projected in this part of the diagram were low capelin years associated with high growth rates of capelin (1988–1989, 1994–1998, 2004–2006).

Figure 2.2.5. Results from PCA on 23 biotic variables for fish related to stock size, recruitment, maturation, growth, and mortality of different species fish (cod, haddock, capelin, herring, polar cod, saithe, Greenland halibut, and long-rough dab). Biplot of variables and years plotted on PC-1 and PC-2.Figure 2.2.5. Results from PCA on 23 biotic variables for fish related to stock size, recruitment, maturation, growth, and mortality of different species fish (cod, haddock, capelin, herring, polar cod, saithe, Greenland halibut, and long-rough dab). Biplot of variables and years plotted on PC-1 and PC-2.

Figure 2.2.6. Same as Figure 2.2.4 but with environmental (abiotic) variables projected as passive variables onto the biplot.Figure 2.2.6. Same as Figure 2.2.4 but with environmental (abiotic) variables projected as passive variables onto the biplot.

The overall change in the Barents Sea has been from cold conditions, low demersal stocks levels and high fishing pressure in the 1980’s to warm conditions (Chapter 3.1), large demersal stocks (Chapter 3.6) and lowered fishing pressure (Chapter 3.9.1), in recent years (Figure 2.2.7).

Figure 2.2.7. Time-series of annual means of abiotic and biotic variables in the Barents Sea sorted by trend. Blue means value below the 1988-2016 average and red means above the 1988–2016 average.Figure 2.2.7. Time-series of annual means of abiotic and biotic variables in the Barents Sea sorted by trend. Blue means value below the 1988-2016 average and red means above the 1988–2016 average.