Assessing Chinese fishing effort off the Galápagos Marine Reserve: fishRman like Oceana?

Following fishRman‘s submission to the JOSS (Journal of Open-Source Software), a first peer-review has recently come out. The feedback was great and pointed out the need, for the documentation, to present examples on how to solve real-world analysis problems. This is something I had already promised at the beginning of the Handbook, but forgot along the creation process.

Instead of presenting fictitious scenarios to solve, I decided to put fishRman to the test and see how it fared in comparison to a famous report by Oceana. The guide that follows is an excerpt from the fishRman Handbook, which I suggest reading if you want to follow along.


In the second half of 2020, multiple news outlets reported on the Chinese fleet exerting an intense fishing pressure barely off the borders of the Marine Protected Area of the Galápagos Islands. The Guardian, in particular, reported some interesting numbers from a work by Oceana. As an example, the reader will be instructed on how to research this topic and fact-check this piece of news.

Note here that Oceana used advanced techniques to account for fishing activity exerted by vessels with AIS transmitters turned off, while fishRman works solely on ‘visible’ activity. The results will thus slightly differ, while also being equally correct in their context and with their premises.

This guide will fact-check 3 Points of the report by Oceana:

  1. Chinese vessels account for 99% of the fishing effort;
  2. The main target of the Chinese fleet was squid;
  3. Spatial distribution of the fleet (visual comparison of the maps produced).

First of all, research the coordinates of the area to investigate. Data from a broader area than the islands alone should be queried in the ‘Query’ tab, adjusting the query through a process of trial and error until the reader is satisfied with the spatial extent of the analysis. The table to query is the one with AIS data gridded at 0.01 degrees, since it retains information about the flag state of the vessels.

As indicative values, this work suggests a latitude between -9 and 5, and a longitude between -100 and -80. The time period is stated in the sources as from 2020-07-13 to 2020-08-14. This suggested query is more formally written as this SQL Query:

SELECT * FROM `global-fishing-watch.gfw_public_data.fishing_effort_v2` WHERE date >= '2020-07-13' AND date < '2020-08-14' AND cell_ll_lat >= -9 AND cell_ll_lat < 5 AND cell_ll_lon >= -100 AND cell_ll_lon < -80 

The reader is encouraged to find the interval of coordinates that best fits the analysis, bearing in mind that their results will diverge from the ones below if they do so.

When fishRman is done fishing the data, it is time to leave the ‘Query’ tab in favor of the ‘Analysis’ tab. Here, the focus is on the ‘Descriptive’ tab, in the ‘Available analyses’ section of the sidebar. Since the query encompasses multiple months, a way to calculate the total fishing hours exerted is to aggregate by ‘year’ by checking the relative checkbox and hitting the ‘Summarize’ button. The analysis returns a ‘Total fishing’ value of 79520.49 hours.

Focusing now on the Chinese fleet alone, the reader is invited to uncheck the abovementioned ‘year’ checkbox, checking the ‘flag’ checkbox instead (leaving the ‘year’ box checked returns the same result, it is only unchecked to focus the attention of the reader onto the new topic). This will calculate the total fishing hours exerted by each nation represented in the queried data. The analysis returns a ‘Total fishing’ value for CHN (Chinese) vessels of 70658.1493 hours.

In order to have the fishing effort of the Chinese fleet as a percentage of the total, one just needs to use the simple equation:

(Chinese fishing hours / Total fishing hours) * 100 = (70658.1493 / 79520.49) * 100 = 88.855274%

That amount is in line with the general idea of one fleet dominating the fishing arena, and is accurate and correct with the right premises, even though it is still far from the 99% reported by Oceana.

Readers with minimal knowledge of GIS software, of which QGIS is the most renowned Open-Source representative, might take this analysis a step further. In particular, they might download from marineregions the shapefile for the Exclusive Economic Zone (EEZ) of the Galápagos islands, the borders of which correspond to the Marine Protected Area currently being analysed, create a buffer area outside the borders, and export the layer as a GeoPackage. This helps ignoring the fishing effort exerted within the EEZ or too far from it and focusing on the one exerted immediately out of it. For this guide, a buffer of 2 degrees of latitude-longitude was created. The reader is encouraged to find the buffer distance (width) that best fits the analysis, bearing in mind that their results will diverge from the ones below if they do so.

To use this buffer area, the reader must ‘Convert’ the queried data into spatial data, from the ‘GPKG’ tab, at the top of the sidebar. Then, they must upload the GeoPackage file for the buffer as the ‘Area of interest’, choose the correct layer, and check the ‘Use only data contained in the area?’ checkbox.

Now, they must repeat the ‘Descriptive’ analyses described above, one for ‘year’ and one for ‘flag’, and use the same equation for calculating the percentage. This will bear the following results:

Table 2: Fishing hours by all fleets combined and by the Chinese fleet alone in comparison.

Chinese fishing hours Total fishing hours Chinese/Total percentage
70615.9944 73101.13 96.600414%

As expected, focusing the analyses on the area immediately outside the MPA returns results that are comparable to those published by Oceana.

Point 1 has successfully been checked.

The next point focuses on the Chinese fleet, thus needing the reader to query the data again with the same parameters, this time adding a ‘flag’ filter for Chinese (CHN) vessels. The reader must then open the ‘Analysis’ tab and ‘Convert’ the data into spatial data, from the ‘GPKG’ tab, in the sidebar.

Note here that a researcher must be consistent in their analysis, or explain why they chose not to. The reader must thus ignore, or repeat, the optional buffer area step accordingly.

The reader must then move to the ‘Descriptive’ tab, in the ‘Available analyses’ section of the sidebar. Here, they must check the ‘flag’ checkbox, and click the ‘Summarize’ button. The resulting tables have been merged and the numbers rounded to the first decimal for clarity.

Table 3: Fishing hours and vessel hours for each geartype employed by the Chinese fleet. The numbers are reported before and after filtering by the buffer area of interest, and the difference between the two.

table3

It is clear there is little difference between using or not the buffer area in terms of sheer results. Since they all match, and confirm what was expected to be found in Point 2: Squid jiggers, whose main target species is squid, exerted the majority of the fishing effort for the Chinese fleet in the area.

For the last point, the reader must then move to the ‘Spatial’ tab, in the ‘Available analyses’ section of the sidebar. Here, they must click ‘Visualize’. This guide uses the following parameters:

  • Also show global: EEZ
  • Map by: Total fishing hours
  • Top % of dataframe: 99%
  • -98 < Longitude < -83
  • -7 < Latitude < 5
  • Map resolution: 0.05

Figure 6: Visualization by Oceana (left) and fishRman (right) in comparison.

The spatial distributions of the fleet match in detail, with the same shapes, hotspots, and empty areas being reported.

The fact-check of Point 3 ends this example.

The reader is now capable of performing analyses and delivering visualizations that are comparable for accuracy, precision, and truth to facts, to those present in the report by Oceana.

(Featured image by Ryoga Otake on Unsplash)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s