Estimating relative abundance from catch and effort data, using neural networks

We develop and test a method to estimate relative abundance from catch and effort data using neural networks. Most stock assessment models use time series of relative abundance as their major source of information on abundance levels. These time series of relative abundance are frequently derived from catch-per-unit-of-effort (CPUE) data, using general linearized models (GLMs). GLMs are used to attempt to remove variation in CPUE that is not related to the abundance of the population. However, GLMs are restricted in the types of relationships between the CPUE and the explanatory variables. An alternative approach is to use structural models based on scientific understanding to develop complex non-linear relationships between CPUE and the explanatory variables. Unfortunately, the scientific understanding required to develop these models may not be available. In contrast to structural models, neural networks uses the data to estimate the structure of the non-linear relationship between CPUE and the explanatory variables. Therefore neural networks may provide a better alternative when the structure of the relationship is uncertain. We use simulated data based on a habitat based-method to test the neural network approach and to compare it to the GLM approach. Cross validation and simulation tests show that the neural network performed better than nominal effort and the GLM approach. However, the improvement over GLMs is not substantial. We applied the neural network model to CPUE data for bigeye tuna (Thunnus obesus) in the Pacific Ocean.

Exploring the use of neural networks for biomass forecasts in the Peruvian upwelling ecosystem

A pilot study was conducted to study the ability of an artificial neural network to predict the biomass of Peruvian anchoveta Engraulis ringens, given time series of earlier biomasses, and of environmental parameters (ocenographic data and predator abundances). Acceptable predictions of three months or more appear feasible after thorough scrutiny of the input data set.

ROV-based Deep Water Monitoring of the Northern Channel Islands Marine Protected Areas Annual Report - 2007

Research cruises were conducted in August-October 2007 to complete the third annual remotely operated vehicle (ROV)-based assessments of nearshore rocky bottom finfish at ten sites in the northern Channel Islands. Annual surveys at the Channel Islands have been conducted since 2004 at four sites and were expanded to ten sites in 2005 to monitor potential marine protected area (MPA)effects on baseline fish density. Six of the ten sites are in MPAs and four in nearby fished reference areas. In 2007 the amount of soft-only substrate on the 141 track lines surveyed was again estimated in real-time in order to target rocky bottom habitat. These real-time estimates of hard and mixed substrate for all ten sites averaged 57%, 1% more than the post-processed average of 56%. Surveys generated 69.9 km of usable video for use in finfish density calculations, with target rocky bottom habitat accounting for 56% (39.1 km) for all sites combined. The amount of rocky habitat sampled by site averaged 3.8 km and ranged from 3.3 km sampled at South Point, a State Marine Reserve (SMR) off Santa Rosa Island, to 4.7 km at Anacapa Island SMR. A sampling goal of 75 transects at all 10 sites was met using real-time habitat estimates combined with precautionary over-sampling by 10%. A total of seventy kilometers of sampling is projected to produce at least seventy-five 100 m2 transects per site. Thirteen of 26 finfish taxa observed were selected for quantitative evaluation over the time series based on a minimum criterion of abundance (0.05/100 m2). Ten of these 13 finfish appear to be more abundant at the state marine reserves relative to fished areas when densities were averaged across the 2005 to 2007 period. One of the species that appears to be more abundant in fished areas was señorita, a relatively small prey species that is not a commercial or recreational target. (PDF contains 83 pages.)