7.1 Predicting Suitable Catches

Author

Clyde Blanco

Published

March 26, 2025

📣 Attention: We will utilize R as the programming language for data processing and model training.

The electronic logbook of skippers registers information on the fishing operations such as fish landings and sales. This mandatory recording of catches has the potential usage in building models such as species distribution models due to the frequency and coverage of data collection. However, this data are not usually used for such kind of modelling because the resolution is coarse (i.e., per ICES rectangle) and are more prone to erroneous records. The resolution and quality of catch data coming from VISTools opens up the possibility to develop better models.

Fig 1. Diagram showing the step by step process in building the Fishing Suitability Model from data collection until model deployment.

Fig 1. Diagram showing the step by step process in building the Fishing Suitability Model from data collection until model deployment.

The previous section outlines our objective to create models that facilitate informed decision-making, including the Fishing Suitability Model (Fig 1). This model focuses on two key species for the Belgian fleet: common sole (Solea solea) and European plaice (Pleuronectes platessa). It generates forecast maps that pinpoint suitable areas for catching these species, promoting selective and sustainable fishing practices. Catches are classified as either “suitable” or “not suitable” based on a defined threshold (Vilela & Bellido, 2015). We have set these thresholds at 50 kg for sole and 150 kg for plaice, based from interviews from vessel owners. These thresholds are then standardized against the mean fishing effort, measured in haul duration (hours). To compute the mean effort, we retain fishing hauls that meet or exceed the threshold weights for each species and then average the fishing effort. For brevity, we will refer to the fish species using FAO codes: SOL for sole and PLE for plaice in this section and the following ones.

SOL_mean_effort <- catch_data %>%
  filter(Species=="SOL") %>%                # retain hauls with sole
  filter(Weight>=50) %>%                    # filter out hauls with less than 50 kg of sole
  drop_na(HaulDur) %>%                      # remove rows with missing haul duration 
  summarise(mean = mean(HaulDur))           # compute mean fishing effort

SOL_thres <- 50/(SOL_mean_effort$mean)      # calculate threshold (Catch Per Unit Effort)


PLE_mean_effort <- catch_data %>%
  filter(Species=="PLE") %>%                # retain hauls with plaice
  filter(Weight>=150) %>%                   # filter out hauls with less than 50 kg of plaice
  drop_na(HaulDur) %>%                      # remove rows with missing haul duration 
  summarise(mean = mean(HaulDur))           # compute mean fishing effort

PLE_thres <- 150/(PLE_mean_effort$mean)     # calculate threshold (Catch Per Unit Effort)

Other sources of data can also be pooled to the existing data to improve the prediction accuracy of the model. Data coming from fishery-dependent sampling programs can also be fused to the VISTools data. Only hauls (landings) with the metier TBB_DEF_70-99 (beam trawl, demersal fish, mesh size 70-99 mm) were used during the model training to remove the possible effect caused by the different selectivity of gear types. Before we start coupling the catch data to different environmental parameters in the next section, make sure that your catch data contains the variables indicated in the table below. You can keep CPUE in your data if you want to experiment with CPUE as the response variable at a later stage.

ID Date Species Catch Longitude Latitude
18475 2019-01-27 SOL 0 -0.182410 50.38121
18445 2019-01-25 SOL 0 -0.162253 50.39500
28877 2023-03-02 SOL 0 -0.027013 50.41076
19460 2019-07-01 SOL 0 1.092793 50.52419
1451 2007-01-27 SOL 0 -0.350000 50.43333
13884 2015-05-12 SOL 0 -6.383333 51.71667
5470 2009-07-29 SOL 0 0.883333 50.21667
29878 2023-04-16 SOL 0 -7.578722 51.52049
7521 2010-10-10 SOL 1 2.500000 52.56667
20593 2019-11-08 SOL 1 -6.316450 51.53185





References:

Vilela, R., & Bellido, J. M. (2015). Fishing suitability maps: helping fishermen reduce discards. Canadian Journal of Fisheries and Aquatic Sciences, 72(8), 1191–1201. https://doi.org/10.1139/cjfas-2013-0522

Icons:

Plus icon created by Picons - Flaticon; Select icon created by Freepik - Flaticon; Music equalizer icon created by Creative Stall Premium - Flaticon; 3d model icon created by Freepik - Flaticon; Kg icon created by Pixel perfect - Flaticon Binary code icon created by Freepik - Flaticon; Flounder icon created by Smashicons - Flaticon; Computer icon created by Freepik - Flaticon