Fishing effort data, fishing fleet segmentation, and statistical details used in the expert knowledge elicitation experiment
Scacco, Umberto (2023), Fishing effort data, fishing fleet segmentation, and statistical details used in the expert knowledge elicitation experiment, Dryad, Dataset, https://doi.org/10.5061/dryad.8cz8w9gw4
Based on an explorative but rigorous elicitation framework, we obtained the bycatch fishing probability at the fishing fleet segment level using expert estimates. Based on the knowledge of three scientific experts, we developed a new and creative structured method for smart and fast fishery-related risk assessments for species of high conservation concern. In order to test the method here propose, we applied it to 76 cartilaginous fish species (included in the IUCN Red Lists) and on five different fishing segments at both Italian and Mediterranean scale. The method produced qualitative results specific to the threat posed by fishing for each species and each segment with information between and within the segments. Based on the interpretation of resilience-disturbance interactions developed for ecological systems, the quantitative results provided reliable cumulative metrics, measuring the extinction risk due to fishing and the response to overfishing for the species considered. Additionally, the results highlight that the method performs best on a small geographic scale. Therefore, the application of this new method on other subregional or local scales where very few data are available (e.g. fishing effort) could be a valuable tool for the preliminary assessment for species of conservation concern. In fact, despite the absence of detailed catch data at local geographic scales, the flexibility of this method could help to highlight potential fishery-related conservation problems and thus redirect conservation strategies for threatened marine species such as many sharks and rays species.
The present dataset contains two different parts, hereafter reported as (1) and (2). (1) is about fishing effort data and fleet segmentation we used to implement the metrics used in the method for species risk assessment to fishing. Fishing effort data cover a four-year period and are related to Mediterranean and Italian scales. (2) is about the statistical method used to perform the elicitation of the expert knowledge of three experts. In particular, Beta distributions are provided according to species, expert and fishing segment.
(1) Fishing effort data were downloaded at the publicly available JRC database (JRC, 2020, downloaded at DCF website https://datacollection.jrc.ec.europa.eu/data-calls) referring to a four-year period (2015–2018). The database contains records at the scale of a single vessel, together with some fishery information, provided according to the official standard codification (JRC, 2020). The data considered for analysis were those referred to vessels censed in Mediterranean countries, intended as data on member state fishing activities carried out in Mediterranean and Black Sea waters. Records selected for the analysis were those having the following information at least: 1) macro and sub-areas, 2) vessel belonging country, 3) fishing effort by year (expressed as gross tonnage x fishing days by year), 4) vessel length (expressed as categories based on length size ranges), 5) category of fishing gears yearly used and expressed as official identification alphanumerical code (FAO, 2016) and 6) a valid fishing effort data entry. For gear codes and a short description of the different gear please visit http://www.fao.org/3/a-bt988e.pdf.
Based on database information, we extracted and grouped data according to the three criteria in the opinion of the authors:
a) Ensuring a more focused representation of the threat at the species level extracting effort data for fishing gear that potentially poses a threat to cartilaginous species ► selecting fishing gear.
b) Providing a better representation of fishing effort in the two main fishing areas (i.e. inshore and offshore) ► grouping effort data by vessel length.
c) Better representation of the selectivity of the fishing gear towards the considered species, as an ‘average’ selectivity at the level of the fishing segment, ► grouping gear effort data by representative fishing segment.
Fishing effort data are divided into two CSV files named “Selected Fishing Effort Data at the Mediterranean scale” and “Selected Fishing Effort Data at the Italian scale”. The two CSV files contain the quantitative effort data considered and selected in the analyses at the Mediterranean and Italian scales, respectively. We used total gross tonnage (GT) x fishing days (fishing days) (contained in the field “total GT fishing days”, in column B of each file) as the best measure for fishing effort because such a product takes into consideration both the engine power of the boat and corresponding fishing time actually spent at sea. The mean fishing effort by fishing segment was calculated on the individual records contained in these two files. The mean fishing effort was subsequently used to calculate quantitative metrics of the species risk assessment to fishing. This information allows for 1) the independent check of the results presented in this work; 2) the replication of the method used to extract, select and use fishing effort data at different scales (see also 2.3 in the main paper).
(2) According to the methods reported in the main paper, the elicitation framework of experts ‘estimates is provided in the following CSV files: " Elicitation framework, Mother Distribution Models of experts ‘estimates ", " Elicitation framework of experts ‘estimates, Beta Fitted Distributions", and " Elicitation framework of experts ‘estimates, Linear Combinations of quartiles". The first file contains statistics of the distributions we used to obtain an upper and lower bound for the single-value estimates that each expert provided for each fishing segment in the elicitation experiment (see 2.1 in the main paper for reference). The second file reports statistics of the Beta fitted distributions we used to describe the variation range of estimates by species and by expert. Beta distributions are usually considered the best models to fit estimates provided using expert knowledge (see 2.1 and Appendix A 3 in the main paper for reference). Finally, the third file shows the linear combination of quartiles from pooled estimates of the experts by species and fishing segment. These data were subsequently used for the species risk assessment for fishing (see also 2.2 in the main paper).
This information allows for 1) the independent check of the results presented in this work; 2) the replication of the method used to elicit least expert knowledge for multiple dependent quantities that sum up to 1 (see also 2.1 in the main paper).
The data can be opened through Excel software