4 Results

4.1 Coverage

4.1.1 Geographical coverage

Should probably include a map here of the countries that we have included. Here is a list instead to start with

A complete set of trait and Aquamap data (allowing the calculation of climate hazard) was gathered for 147 species caught in European waters, representing 90.3% of the total value of fish and shellfish landings in Europe. Catch data was available covering 26 countries in Europe: country-wise coverage was at least 78% by value and typically above 90%.

When combined with catch data covering 23 FAO subareas, this resulted in a total of 1300 potential “stocks”. However, many of these stocks made relatively minor contributions to the total catch of the species: stocks comprising less than 5% of the total catch of that species were therefore filtered out, leaving 523 stocks in the analysis.

4.1.2 Fleet Coverage

We should also check the coverage by fleet segment. We are able to do the full analysis for 404 fleet segments, although the coverage of some of these units can be pretty poor.

Nevertheless, we still cover 75% or more of the landings value for more than 70% of fleets. We should probably filter this at some point to focus on the most important and/or best covered fleets.

Number of fleet segements: 404

4.2 Biological metrics

4.2.1 Species-specific metrics

As a cross check that we have implemented the coversion from these metrics to a hazard, we plot one against the other, to ensure that the trend is as intended

So, longer-lived species have a higher hazard score. Tick.

So species with a higher habitat-specificity have a higher hazard. Tick.

4.2.2 Stock-specific metrics

Small stocks are a potential problem in this analysis. We first check the distribution of stock sizes (expressed as value of landings) relative to the total size (value) of the species.

We can therefore see that there are an awful lot of stocks that don’t contribute much landigns value - is is a natural consequence of the way that we have defined a stock in this case. We choose to filter out all stocks that have a landings value less than 5% of the species total.

This leaves us with 523 stocks. Looking at the distribution of TSM’s of these remaining stocks.

And cross checking the hazard implementation.

Low TSMs => higher hazards. Tick. The stripping rises due to the categorical natural of the other metrics.

4.2.3 Species-level hazard

We can look at our biological analysis by integrating the hazard back up to the species level by weighting by the value of landings of each stock.

Looking at the top and bottom ten species

FAO_3A_code FAO_EN hazard rank
WHE Whelk 0.9322289 1
DEC Common dentex 0.9062602 2
SCE Great Atlantic scallop 0.8255320 3
KLK Smooth callista 0.8255122 4
MEG Megrim 0.7937851 5
OYF European flat oyster 0.7916300 6
LEZ Megrims nei 0.7835960 7
FOR Forkbeard 0.7636658 8
RSE Red scorpionfish 0.7396029 9
SVE Striped venus 0.7231577 10
WHB Blue whiting(=Poutassou) 0.2694861 139
CET Wedge sole 0.2589207 140
SQI Northern shortfin squid 0.2563509 141
HER Atlantic herring 0.2481068 142
OCZ Octopuses nei 0.2451900 143
SQM Broadtail shortfin squid 0.2419941 144
SQC Common squids nei 0.2402267 145
BLT Bullet tuna 0.2056274 146
BON Atlantic bonito 0.1534985 147
SSH Scarlet shrimp 0.1232536 148

4.3 Fleet metrics

4.3.1 Fleet hazard metrics

Fleets with the top and bottom 10 hazard scores

Table 4.1: Top and botton 10 fleets by hazard
fs_name value.covered total.val.landings prop.missing hazard
PRT A27 DTS1824 45668.0 425510.0 0.8926747 0.2150344
FIN A27 TM1218 ° 10807913.8 10983806.6 0.0160138 0.2469837
EST A27 PG1012 7613725.0 7630744.0 0.0022303 0.2479600
FIN A27 TM1824 30019850.0 30019850.0 0.0000000 0.2481674
FIN A27 TM2440 ° 107055789.0 107065174.7 0.0000877 0.2483668
EST A27 TM1218 860437.0 860437.0 0.0000000 0.2542685
EST A27 TM2440 ° 48072014.0 48072264.0 0.0000052 0.2564641
LTU A27 TM2440 ° 13475876.0 13475876.0 0.0000000 0.2609496
POL A27 TM2440 ° 73972680.3 75163367.3 0.0158413 0.2657755
LVA A27 TM1218 7448064.0 7711364.0 0.0341444 0.2667060
BEL A27 PMP1824 ° 473171.6 495884.6 0.0458029 0.7841981
GBR A27 DRB0010 2022678.0 2657277.3 0.2388156 0.7845295
ITA A37 DRB1218 ° 17394834.4 17772573.7 0.0212541 0.7892047
FRA A27 DRB0010 1406833.8 8547518.7 0.8354103 0.7907565
GBR A27 DRB1012 1972195.2 2077496.5 0.0506866 0.8141036
FRA A27 FPO0010 7708186.0 8200296.5 0.0600113 0.8192936
ITA OFR DTS40XX IWE 2955.0 370659.0 0.9920277 0.8281156
IRL A27 FPO0010 4408690.9 5406757.3 0.1845961 0.8318423
IRL A27 DRB2440 ° 2004905.0 2004905.0 0.0000000 0.8374500
FRA A27 FPO1012 8163528.9 8435296.8 0.0322179 0.8576904
Distribution of hazard metrics

Figure 4.1: Distribution of hazard metrics

4.3.2 Fleet exposure metrics

We use the Shannon diversity and the Simpson Dominance metrics to characterise the diversity of species that a fishing fleet catching. The distribution of these metrics is as follows.

Fleetwsie Shannon Diversity index

Figure 4.2: Fleetwsie Shannon Diversity index

Fleetwise Simpson Dominance

Figure 4.3: Fleetwise Simpson Dominance

A correlation between these two metrics is expected.

Relationship between Shannon Diversity H’ and Simpson’s Dominance D of fishery landings, for EU fleet segments in 2016

Figure 4.4: Relationship between Shannon Diversity H’ and Simpson’s Dominance D of fishery landings, for EU fleet segments in 2016

Clearly there is a strong relationship between Shannon diversity of fishery landings and Simpson’s dominance (D), although this is not linear (figure 1).

We combine the two metrics together into a into exposure metrics. Checking that we have done this correctly:

Exposure vs Shannon Index

Figure 4.5: Exposure vs Shannon Index

So fleets with higher diversity have lower exposure. Tick. And fleets that are dominated by a few species have a higher exposure, which is what we want.

Plotting the list of fleet diversity.

Least diverse fleets

Table 4.2: Least diverse fleets
fs_name sum.prop sum.value shannon.H simpsons.D n.species exposure
IRL A27 DRB2440 ° 1 2004905 0.0000000 1.0000000 1 1.0000000
MLT A37 PS2440 1 360500 0.0000000 1.0000000 1 1.0000000
PRT A37 FPO2440 1 116207 0.0000000 1.0000000 1 1.0000000
ESP A27 DRB1218 1 2285575 0.0049305 0.9989311 3 0.9988634
DEU A27 TBB1012 ° 1 61863 0.0058449 0.9986430 3 0.9986065
ROU A37 PMP1218 ° 1 4157697 0.0214070 0.9944113 10 0.9945785

Most diverse fleets

Table 4.3: Most diverse fleets
fs_name sum.prop sum.value shannon.H simpsons.D n.species exposure
ESP A37 PMP0612 1 4814591 3.743704 0.0589389 347 0.0705235
FRA A37 DFN0612 1 1337565 3.744439 0.0373679 203 0.0593585
ESP A37 DTS2440 1 5647283 3.841203 0.0431922 476 0.0508161
ITA A37 PGP0612 1 21059018 3.854408 0.0363553 135 0.0457311
ESP A37 DTS1218 1 4437824 4.033915 0.0326257 701 0.0224204
ESP A37 DTS1824 1 11824853 4.194966 0.0263463 746 0.0000000

Catch diversity in 2016 ranged from zero (where a fleet caught a single resource) to 4.19, where a multitude of different fish and shellfish species were targeted. The lowest diversity of catches was observed for the fleet segments IRL-A27-DRB2440, MLT-A37-PS2440 and PRT-A37-FPO2440, fishing exclusively on Great Atlantic scallop (Pecten maximus), Chub mackerel (Scomber colias) and Striped soldier shrimp (Plesionika edwardsii) respectively. By contrast, the highest diversity of landings was observed for ESP-A37-DTS1824, with catch records for 746 different species.

4.3.3 Fleet vulnerability metrics

Visualisation of the distribution of metrics
Employment per vessel

Figure 4.6: Employment per vessel

This is a potentially problematic metric, in retrospect, as it is really a metric of vessel size…

QUESTION: Should we retain it? How should we normalise it if we keep it?

Value of landings

Figure 4.7: Value of landings

Hmmm. Similar problem.

Average wage per FTE

Figure 4.8: Average wage per FTE

Looks like a couple of outliers there that need to be polished up…

Net profit margin

Figure 4.9: Net profit margin

Hmmm. A rather biased distribution. Lets crunch this figure down a bit…

Net profit margin 2

Figure 4.10: Net profit margin 2

Outliers aside, this is not a bad distribution.

QUESTION: Shall we rescale it to +/-50, and trim the rest…?

For the sake of this analysis, we have only based vulnerability on the Net Profit Margin, while we decide whether to include the rest..

Looking at the total vulnerability distribution.

Vulnerability distribution

Figure 4.11: Vulnerability distribution

This doesn’t look so bad, but at the same time the outliers probably don’t help things, and should be fixed.

Checking the correlations with the metrics against vulnerability
Correlation check

Figure 4.12: Correlation check

Which highlights the problems quite nicely… There is also some issues here with the directionality that we need to fix.

I think we probably need to rethink these metrics. Net profit margin is a very useful metric, but I’m not so sure about the others, particularly as they scale with vessel size, which I’m not sure we realy want…

4.3.4 Fleet risk

Note that the results here are a bit meaningless until we get the vulnerability under control, but the visualisations will be the same, so we can look at and discuss how we want to approach these.

Firstly, the top and bottom 10

Table 4.4: Top and bottom 10 fleets by risk
fs_name total.val.landings prop.missing hazard exposure vulnerability risk
ITA A37 PGP0006 5721461.6 0.459 0.464 0.019 0.038 0.001
ESP A37 DFN0612 630328.7 0.477 0.469 0.024 0.073 0.004
DNK A27 TM40XX 363359393.3 0.007 0.048 0.533 0.019 0.006
PRT A27 DFN1012 356623.0 0.390 0.457 0.033 0.137 0.009
PRT A27 DTS0010 381875.0 0.275 0.484 0.150 0.024 0.011
ESP A37 DTS0612 277330.1 0.627 0.434 0.075 0.150 0.014
ESP A37 DRB1218 167984.4 0.460 0.127 0.397 0.189 0.016
ITA A37 PGP0612 21059018.0 0.454 0.422 0.006 0.288 0.019
EST A27 PG0010 3960796.0 0.538 0.123 0.447 0.152 0.021
ITA A37 DTS1218 23840875.4 0.261 0.486 0.041 0.197 0.024
BEL A27 PMP1824 ° 495884.6 0.046 0.976 0.880 0.623 0.976
FRA A27 MGP0010 ° 3267734.5 0.942 0.645 0.925 0.910 0.979
PRT OFR HOK2440 IWE° 4408259.0 0.766 0.890 0.595 0.999 0.981
GBR A27 HOK1012 ° 355038.7 0.082 0.969 0.858 0.704 0.984
IRL A27 FPO0010 5406757.3 0.185 0.994 0.580 0.959 0.986
POL OFR TM40XX 53086199.0 0.635 0.835 0.771 0.989 0.989
IRL A27 FPO1012 5106661.0 0.072 0.954 0.684 0.962 0.991
ESP A27 DRB1218 2285575.2 0.001 0.952 0.996 0.660 0.994
IRL A27 FPO1218 ° 5015581.4 0.166 0.887 0.778 0.964 0.996
IRL A27 DRB2440 ° 2004905.0 0.000 0.996 0.999 0.944 0.999

Now, viewing by vessel size

By country and fishing area.

QUESTION: I’m not quite sure how whether to retain the other fishing region fleets, or just keep it to the FAO27 and FAO37 areas - we don’t have very much biological information outside these regions.

QUESTION: I can also plot this as a map, based on the median value…

Fleetwise climate risk for the 23 EU coastal nations

Figure 4.13: Fleetwise climate risk for the 23 EU coastal nations

TODO: Ireland needs to be investigated - it looks like it is missing some of the flet economic data => no vul => no risk.

TODO: I will separate France and Spain into Atlantic and Mediterranean fleets as well on this map

And by gear. Looks like quite some variability here.

A table of gear types, sorted by median risk.

Code Description Risk
DRB Dredgers 0.877
PGO Vessels using other passive gears 0.741
FPO Vessels using pots and/or traps 0.679
MGP Vessels using polyvalent active gears only 0.662
TBB Beam trawlers 0.575
PMP Vessels using active and passive gears 0.515
MGO Vessel using other active gears 0.496
DFN Drift and/or fixed netters 0.472
TM Pelagic trawlers 0.442
HOK Vessels using hooks 0.434
PGP Vessels using polyvalent passive gears only 0.417
DTS Demersal trawlers and/or demersal seiners 0.400
PG Passive Gears 0.311
PS Purse seiners 0.306

4.4 Regional Metrics

4.4.1 Regional hazard

4.4.2 Regional exposure

Similar relationships between the catch dominance and catch diversity are seen at the regional levels as are seen for the fleets.

Check on the scaling and tendency

Plotting the exposure score geographically

Note the clear north-south gradient here. Several countries are still missing from this analysis, as we don’t have data resolved by NUTS regions for them (e.g. Norway, Finland, parts of Sweden, SE Europe). We need to find a strategy how to deal with this.

4.4.3 Regional vulnerability

Currently this is only based on GDP per capita. Other metrics can be added. Plotting the distribution of values.

Checking the scaling and directionality

Distribution map.

4.4.4 Regional risk

We calculate risk as the average of the hazard, exposure and vulnerability scores. This may need to be improved. The top and bottom ten regions by risk are then

Table 4.5: Top and bottom 10 regions by risk
region hazard exposure vulnerability risk country
ITH5 0.123 0.221 0.191 0.005 IT
NO 0.113 0.456 0.054 0.015 NO
NL32 0.172 0.534 0.034 0.025 NL
DK05 0.250 0.407 0.123 0.034 DK
SE 0.103 0.642 0.074 0.044 SE
FRG0 0.368 0.083 0.377 0.054 FR
FRE1 0.181 0.113 0.554 0.064 FR
BE25 0.221 0.466 0.181 0.074 BE
ITI4 0.534 0.054 0.289 0.088 IT
UKM5 0.466 0.368 0.044 0.088 UK
UKF3 0.544 0.907 0.623 0.907 UK
UKD1 0.995 0.838 0.260 0.917 UK
UKC1 0.956 0.554 0.613 0.926 UK
UKE4 0.877 0.926 0.328 0.936 UK
UKE1 0.887 0.799 0.505 0.946 UK
UKM8 0.966 0.975 0.270 0.956 UK
UKM9 0.985 0.652 0.701 0.966 UK
RO 0.583 0.877 0.975 0.975 RO
IS 0.564 0.995 0.995 0.990 IS
LT02 0.779 0.848 0.926 0.990 LT

Putting it on a map

Putting it up as a boxplot

Interesting results. Lots of details there to discuss!