CAGUA et al_revised_20140418.pdf (1.17 MB)

Acoustic Monitoring of Blast Fishing: Pilot Study - Dar es Salaam

Download (0 kB)
Version 2 2014-04-21, 23:39
Version 1 2014-04-21, 01:49
journal contribution
posted on 2014-04-21, 01:49 authored by Edgar fernando CaguaEdgar fernando Cagua, Christopher Basset, Robert Kayanda, Jason Rubens

Technical report: WWF Tanzania Country Office - Marine Programme

Blast fishing is illegal fishing method according to Fisheries Act (2003) due to its negative impact especially on coral reefs. The method is widely used and it is a serious problem along the coast of Tanzania mainland. The most affected area is Dar es Salaam, where existing regulations are not effectively enforced to address the problem. Inadequate data on the frequency of occurrence and distribution of blast activity prevent effective control measures.

Using Digital Spectrogram Long-Term Acoustic Recorders (DSG), we performed two deployments. One was a two-day long deployment to identify the acoustic characteristics of blasts, and the second was a two- month long deployment to monitor blast activity off Dar es Salaam coast.

In the blast identification phase we deployed one DSG over two days on Milliards Bank recording sound at 80 kHz, we analysed the acoustic characteristics of 45 underwater explosions to identify and characterize blasts used during illegal fishing operations. We developed a neural- network that semi-automatically identifies explosion-like recordings. Ninety-five percent of the explosion-like events identified by the neural network are attributable to blast fishing, while the other five percent were other unidentified transient signals.

During the monitoring phase two DSGs were deployed, however we were able to retrieve only one of them. Weather, currents, and possibly vandalism might be the causes for the loss. The retrieved DSG was deployed on Mbudya patches between 15 April and 30 May, 2014 recording at a sampling frequency of 40kHz during five minutes every ten minutes between 0600 and 1800 hrs. Using this additional data the neural network improved its classification accuracy to 98%. We estimate an average of 19 blasts per day during daytime off Mbudya, with most of the blasts occurring in the morning before 1300 hrs.

To avoid further loss of instruments, we recommend the use of acoustic releases in further deployments. However care must be taken to avoid potential problems caused by biofouling as well as preventing interference from noise near the DSGs caused by floats and moving elements part of the release.

In order to better understand blast fishing patterns as well as factors influencing underwater noise production we also recommend recording environmental variables related to weather and currents.

This work represents the first steps of a potential future monitoring program off the Tanzanian coast. We provide methods and recommendations for future monitoring of blast fishing



Usage metrics



    Ref. manager